Raw Transcript 1-877-FACTSET www.callstreet.com Total Pages: 36 Copyright © 2001-2018 FactSet CallStreet, LLC 01-Mar-2018 Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 2 Copyright © 2001-2018 FactSet CallStreet, LLC CORPORATE PARTICIPANTS Scott D. Wollney President, Chief Executive Officer & Director, Atlas Financial Holdings, Inc. Paul Anthony Romano Chief Financial Officer & Vice President, Atlas Financial Holdings, Inc. ...................................................................................................................................................................................................................................................... OTHER PARTICIPANTS Jon Paul Newsome Analyst, Sandler O'Neill & Partners LP Matthew J. Carletti Analyst, JMP Securities LLC William Joseph Dezellem Chief Investment Officer & President, Tieton Capital Management LLC Robert Farnam Analyst, Boenning & Scattergood, Inc. Brian Hollenden Analyst, Sidoti & Co. LLC Ronald David Bobman Analyst, Capital Returns Management LLC ...................................................................................................................................................................................................................................................... MANAGEMENT DISCUSSION SECTION Operator: Greetings and welcome to Atlas Financial Holdings Announces Preliminary 2017 Fourth Quarter and Year-End Financial Results. At this time, all participants are in a listen-only mode. A brief question-and-answer session will follow the formal presentation. [Operator Instructions] As a reminder, this conference is being recorded. I would now like to turn the conference over to Scott Wollney, Chief Executive Officer. Thank you. Please begin. ...................................................................................................................................................................................................................................................... Scott D. Wollney President, Chief Executive Officer & Director, Atlas Financial Holdings, Inc. Thank you very much, Rhea, and hello everyone. With me today is Paul Romano, our Vice President and CFO; and Joe Shugrue, our Vice President of Claims. I'll turn it over to Paul to review our policy regarding forward-looking statements before my formal remarks. ...................................................................................................................................................................................................................................................... Paul Anthony Romano Chief Financial Officer & Vice President, Atlas Financial Holdings, Inc. Thank you, Scott, Today, after market close, Atlas issued preliminary financial results for the fourth quarter and year ended December 31, 2017. Copies of this press release are available at the Investor Relations section at the company's website at www.atlas-fin.com. On this call, Atlas may make forward-looking statements regarding the company, its subsidiaries and businesses. Such statements are based on the current expectations of the management of each entity. The words anticipate,

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 3 Copyright © 2001-2018 FactSet CallStreet, LLC expect, believe, may, should, estimate, project, outlook, forecast or similar words are used to identify such forward-looking information. The forward-looking events and circumstances discussed on this call may not occur and could differ materially as a result of known and unknown risk factors and uncertainties affecting the companies, including risks regarding the insurance industry, economic factors and the equity markets generally and the risks factors discussed in the Risk Factor section of its Form 10-K for the year ended December 31, 2016. No forward-looking statement can be guaranteed. Except as required by applicable securities laws, forward- looking statements speak only as of the date on which they are made and the company and its subsidiaries undertake no obligation to publicly update or revise any forward-looking statement, whether as a result of new information, future events or otherwise. When discussing our business operations, we may use certain terms of our, which are not defined under U.S. GAAP. In the event of any unintentional difference between the presentation materials and our GAAP results, investors should rely on the financial information in our public filings. All amounts discussed on this call are in U.S. dollars unless otherwise indicated. We will be utilizing a slide show presentation in conjunction with this call. Though we may address a few slides specifically, in general, we will use this as an accompaniment. With that, I'd now like to turn this call back to Scott. ...................................................................................................................................................................................................................................................... Scott D. Wollney President, Chief Executive Officer & Director, Atlas Financial Holdings, Inc. Thank you, Paul. Last year, we increased reserves to the specific situations, including Michigan and older accident years for global liberty, but had not yet seen broader issues that we now understand are impacting our book. We now recognize the need to take additional steps to address these broader issues. Earlier today, Atlas issued a press release with preliminary financial results for the year ended December 31, 2017 which included reserve strengthening. On this call we'll discuss the reserve changes the timing the income and book value impact and the steps we've taken to address it. We're also taking additional steps to provide greater clarity to all of our investors and partners on a regular basis. Management and the board of directors own approximately 13.2% of Atlas's outstanding shares. This impacts all of us as. Is one of the largest individual owners of Atlas stock I not only take this seriously, but also personally as do all the members of our team. The bottom line is that we're all focused on addressing the issues from the past and improving results for the future. Here are the facts, last year we identified that claims expense in Michigan were significantly outpacing other states and took a significant charge. Today the disproportional amount of claims we've paid in Michigan exceeds $38 million. We also recognize that claims of Global Liberty related to prior years business continued being settled for more than case reserves, while New York business at Global is moving in the right direction particularly not in New York business from 2016 and earlier years developed worse than expected. In addition to these two specific issues while we have reasons to believe that our niche is less exposed to volatility than other areas of commercial auto insurance based on yearend actuarial work there appears to be an overall increase in both paid severity and future expected unpaid amounts across our book of business primarily for accident years 2015 and prior. While we did see positive trends relating to more recent accident years and our predictive model driven business based on year end work we are again moving to our outside independent actuary select point resulting in a significant reserve increase for remaining claims especially those older accident years prior to our use of predictive analytics. The end result is that after the effective reserve strengthening as well as an approximate

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 4 Copyright © 2001-2018 FactSet CallStreet, LLC $0.55 per share DTA write down resulting from the Tax Cut and Jobs Act, book value per share as of December 31, 2017 is expected to be between $7.5 and $8 and statutory surplus across Atlas' four insurance company subsidiaries will be between $85 million and $90 million. Net loss for the full year 2017 was approximately $3.20 per share. From the inception of Atlas, we've consistently increased rates ahead of ISO as shown on slide 4, but we now know that that didn't produce enough premium to generate the kind of margins we expect. While our value proposition to customers was as strong as we've always indicated, claims cost exceeded premium levels particularly in earlier accident years. Starting in 2015, we enhanced our pricing and risk selection with predictive analytics and it has made an impact. Slide 5 illustrates the after-tax impact of reserve changes by program and accident year having moved to the select point developed by our outside actuaries year-end work. At year-end 2017 experienced members of our claims team also conducted a file by file claim review of non-modeled claims as well as all remaining Michigan claims to reinforce our decision making. The results of that review will serve as a benchmark against which sufficiency can be measured as these older claims run off. During the past three years, we've implemented machine learning based predictive analytics tools in both our underwriting and claims operation and included timeline on slide 6 of our deck. Fortunately we are seeing very encouraging evidence that these tools had a marked positive impact and support loss ratios in line with expectations in more recent years. In short, risk selection and pricing precision supported by modeling and underwriting beginning in 2015, appears to have contributed improvement and expected loss ratio for premiums earned in 2016 and 2017. Changes we made in the claims process in mid-2016 were expected to increase the speed with which we identify and pay valid large claims ideally for less than was previously the case and defend non-meritorious or fraudulent claims even earlier with greater efficacy. Case reserves are set to expected ultimate is predicted by our machine learning based modeling tools at 45 days as shown on Slide 7. Typically commercial auto underwriters using traditional methods are not able to establish ultimate case reserves for a year. Deployment of allocated loss adjustment expense dollars are also resulting in better settlements than in the past as seen on Slide 8. While this early data is very encouraging and consistent with expectations, it isn't enough to support a lower reserve charge at this time. We knew these changes would speed claim payments up front and we did see more payments in the near-term during 2017. This included not only more recent claims, but also larger claim payments from the past. The expected compression of Claims Settlement times and payment of model claims for less than case gave us comfort in the overall result until our year-end work was performed. Our changes clearly enabled us to bring potentially larger meritorious claims to ultimate faster. However, until we have additional data showing what we believe will be the positive effects of our predictive analytics, over time we must establish reserves that take into account the possibility that those benefits do not materialize. As a result, as I said our reserves are being strengthened to the select point indicated by our outside independent actuaries. The majority of our case reserves are now based on predictive modeling. Since June 1, 2016 actual payments are less than our underwriting and analytics tools as they would be. We believe the data supports the conclusion that the model is working as intended. We believe the data supports the conclusion that the model is working as intended. We're bringing plans to ultimate faster and more accurately, closing our defending claims in the larger severity ranges as illustrated on slide nine, without increasing average paid severity as shown on slide 10. Going forward, we plan to share quarterly actual losses paid as compared to the estimates from our filed review for older accident years, as well as paid the case reserve outcomes for predictive model-based case reserves for

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 5 Copyright © 2001-2018 FactSet CallStreet, LLC more recent claims. We believe, we've addressed the problem from earlier years head on, further Atlas's core thesis is that as the specialist we should deliver better than industry results and be able to utilize our expertise and sophistication to react to issues surrounding commercial auto more capable than our peers. We implemented the use of analytics and modeling in ways that our competitors have not. And as indicated, we're very comfortable with results from the more recent years. There's clear evidence that in-coming claim volume as well as open claim inventories are both down despite the fact that our book of business grew approximately 20% from 2016 and into 2017 as seen on slide 11. The inflection point highlighted in the chart on the left, shows why we believe we're near the point where the benefits will start to become more evident and our data. As shown on slide 12, our 2017 fully developed accident year loss ratio is coming in as expected at approximately 60%, and our expense ratio is approximately 26% as a result of our operationally efficient platform. We'll continue to leverage analytics to improve relative exposure to risk and should be well-positioned to take additional rate throughout the year. We expect to write in excess of $300 million in premiums in 2018 subject to action by rating agencies. This represents a consistent rate of growth relative to last year. We're comfortable that this continued stable growth can be supported with our existing surplus levels increased use of our reinsurance arrangements and retained earnings going forward. Based on recent commercial auto industry results, we expect further market hardening through 2018. Michigan is materially behind us. We reduced exposure there to less than 1% as seen on slide 13. At this $300 million level of premium and expected use of the company's existing reinsurance programs with a combined ratio in the mid-80s, it's reasonable to expect annual net earnings per share to exceed $2 for 2018. We believe our strategy of providing a specialty product to a unique segment of the commercial auto space remains sound. We also believe based on the data we're seeing that our heavy commitment to an investment in the use of analytics and technology to build on this foundation is paying off. We're applying it more broadly in 2018 and the customers are there and we'll continue to refine our analytics based approach in the years ahead. We'll also capitalize on the favorable commercial auto rate environment. Before we open it up your questions, please know that we're very committed to communicating openly with you in the days, weeks and months ahead. Next week we'll be available for meetings at the Annual [ph] AIFA Insurance Conference and will report our quarterly and annual financial results in fall on March 15. I think most of you know that our executive team is very open to calls and discussions and we expect to provide additional clarity as needed. With that, operator now we're available to take questions.

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 6 Copyright © 2001-2018 FactSet CallStreet, LLC QUESTION AND ANSWER SECTION Operator: Thank you. We will now be conducting a question-and-answer session. [Operator Instructions] One moment please while, we pull for questions. And thank you. Our first question comes from the line of Paul Newsome with Sandler O'Neill. Please proceed. ...................................................................................................................................................................................................................................................... Jon Paul Newsome Analyst, Sandler O'Neill & Partners LP Q Good afternoon. ...................................................................................................................................................................................................................................................... A Hi, Paul. ...................................................................................................................................................................................................................................................... Jon Paul Newsome Analyst, Sandler O'Neill & Partners LP Q Can you go into a little bit more detail about specifically why the case reserves are developing in a negative fashion, is it a – is it lawsuits, is it medical costs, is it – what are the components of that – those case reserved developments? ...................................................................................................................................................................................................................................................... A So we looked at the triangles by line, by state and really other than Michigan continuing to be bad. And for reference on average, the average paid claim in Michigan did not increase a lot in terms of calendar year 2017 to calendar year 2016 payments. They just continue to take up more reserves than would be appropriate based on the amount of premium they generated. And they really isn't a specific hotspot in terms of line of business or state. And in fact, the paid triangles don't really show significant increase in average pays. I think it really was just the aggregate assessment by our outside actuaries in terms of open unpaid claims for 2015 and prior. I wish there was something specific we could point to, but we've reviewed it thoroughly. And again I think it's the broader issues that appear to be – facing commercial auto that we had thought we were more insulated to from those past years. But really when we look at what's left in that inventory, it is a relatively small inventory. And importantly, Joe Shugrue and his team of experienced adjusters with 10 to 20 or more years of experience at the end of the year look at each and every one of those claims and evaluated their expected outcomes. We have a lot of confidence. All of those claims at this point are at least a year old and the amount that the audit the internal audit assessed would be necessary is less than the IBNR we put up for those years. So we do feel we're very comfortable as I've said in the formal comments, again I wish there was a sort of a silver bullet answer. Michigan didn't get better, but it didn't get materially worse and it really was just an assessment of an overall need for a larger IBNR amount to run off the remainder of 2015 and prior based on the work done by our outside actuaries. ...................................................................................................................................................................................................................................................... Q

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 7 Copyright © 2001-2018 FactSet CallStreet, LLC Well, it seems to me there's sort of two components, there's the case reserve itself, which I think usually these are liability claims at the end of the day. If there's a tail to them, so that's usually losses, right. And then the other piece is the IBNR, which is somewhat relating to frequency, but so I guess I was expecting the answer to be simply related to liability losses or court cases, but you're saying that that there's no indication that that's particularly in the issue. ...................................................................................................................................................................................................................................................... A Well, the bulk of it is liability related because you're right, especially for those older years open remaining claims from accident year 2015 and prior are going to be predominantly if not entirely third party liability claims or pip. So if you look at slide 7 in our deck, we show the case reserves that were established as factors. When we put our predictive analytics tool in place is the blue line and the orange line is the case reserves established at 45 days of claim age. And so what it really demonstrates is that for the majority of claims that we have an inventory today, we are estimating expected ultimate at 45 days of claim age and putting that up quickly rather than having a factor reserve and relying on actuarially determined bulk IBNR to make up the difference. And you know again when you look back the bottom line is that for 2015 and prior, the premium levels we charge which were ahead of ISO, they were above average. You know, we have always tried to price on the high-end you know based on the expectation that we're delivering a stronger value proposition. You know, our growth rates were not outrageous. You know, we priced what to what we believe was appropriate based on the information that was available at the time and the losses paid for those years is now outpacing the premium, and as a result you know the conclusion is that the larger reserves are necessary to make sure that there are sufficient to cap off the claims. But you know when we focus on the majority of our inventory, which today are the modeled claims from 2016 and prior. It's a much different situation, because you know we have set the reserves on a claim-by-claim basis using the model, the claims we paid that have been modeled and subsequently closed are our paying for less by a pretty meaningful margin at this point. And as I indicated in the comments you know we have a lot of optimism, that will result in redundancy in those case reserves as we go forward. But because it has only been in-place for about a little over a year and a half, there's not enough data yet to have that support a lower level of case reserves for those more recent years. So it really is two separate issues. There's the runoff of the prior tail where our reserves we relied heavily on the annual year-end IBNR analysis, those claims of the minority of what's left, they've been ring fenced you know by this file-by-file analysis. We're going to be monitoring the closure on those claims very closely, we are committed to disclosing the outcome of those comparisons every quarter And then separately we have the newer claims that we've modeled that have much higher case reserves, and that modeling process as we've talked about is designed to enable us not to have to rely on IBNR once we get pass the end of an accident-year that's got fully reported frequency. And so, it really is two sort of different scenarios. But I guess that I wish there was a more specific issue we could say it is just this line of business in this particular state, but the IBNR allocation is, as you can see on slide 5 is really across all the segments is really just concentrated in accident year 2015 and 2014 in particular. ...................................................................................................................................................................................................................................................... Q Great. Thanks a lot. I'll let other folks ask questions. I appreciate it. ......................................................................................................................................................................................................................................................

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 8 Copyright © 2001-2018 FactSet CallStreet, LLC A Thanks for the questions, Bob. ...................................................................................................................................................................................................................................................... Operator: Thank you. Our next question comes from line Matt Carletti with JMP Securities. Please proceed. ...................................................................................................................................................................................................................................................... Matthew J. Carletti Analyst, JMP Securities LLC Q Hey, thanks. Good afternoon. ...................................................................................................................................................................................................................................................... A Hey, Matt. ...................................................................................................................................................................................................................................................... Matthew J. Carletti Analyst, JMP Securities LLC Q Couple of questions. I guess first Scott, could you walk us through a bit of how the – even it sounds like this year that was a little bit of a more in-depth process but on a more regular basis. How did the quarterly reserve processes take place as compared to say what I presume is a more in-depth Q4 process? Because it seems like this definitely came out caught you guys by surprise that I'm just [indiscernible] that by kind of the interactions that we've had on prior calls and things like that. Just really with the idea of how it seems like this is rather pervasive in the older years, and how it kind of went under the radar for presumably at least three quarters? ...................................................................................................................................................................................................................................................... A Yeah. No, it did catch us by surprise candidly. I mean, we obviously pay close attention to our claim activity as the analytics that we provided in the deck I think illustrate we are looking very closely at claims settlement times, at claim settlement amounts and really in calendar year 2017 there were two things going on simultaneously. There was the run off of the older claims where we were paying above that factor based case reserves and that's what the IBNR we put up at the end of last year should have been there to cover. And at the same time, we were intentionally paying bigger claims faster as they were identified by our predictive modeling. And so when you look at those things in aggregate, the payments for the older claims were using up IBNR, the claims being paid against the predictive model based case. We're releasing redundancy into IBNR and on a quarterly basis it you know as the older claims became the minority and the newer claims became the majority, it looked like those things would essentially offset one another and then. So we knew that the claim payments were accelerating that was by design for the claims predict to be larger and again we've got the specifics in terms of those closure rates on page 9. We were not paying more to close those claims earlier, which is what we illustrate on page 10. So again it wasn't an issue where we saw average paid severity is going up significantly on a calendar year basis. And because those older claims really relied pretty heavily on the IBNR being correct until we did the full year actuarial work. We weren't comfortable coming to a conclusion that on an overall net basis there was a significant need, but we did do a couple of things differently this year obviously than we've done in the past. We did this intensive file review that was possible because at this point claims from 2015 and prior are now at least a year old because

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 9 Copyright © 2001-2018 FactSet CallStreet, LLC almost all the claims for a given accident year get reported within a year. So, we had enough information that the season's adjustors could accurately handicap on a per claim basis what the amount was likely to be. And again, that aggregate number is coming in less than the ultimate amount we strengthened to, but that gives us confidence that it's conservative. And the second thing we did was to look at other outside experts to help us validate some of the things that we think are going to be positives in the future in terms of you know that the pretty fit model based case was being conservative, continuing to refresh all of our modeling based products. And then, the final piece is you know if you look at actual reported frequency, which you know we provide some – some additional detail on, on slide 11, you know our business grew 25% year-over-year from 16% to 17% and our claim inventory on a trailing 12-month basis stayed flat. So, the frequency of new claim coming in versus the cars we were insuring absolutely was going down. And when you look at the chart on the far left that's the inflection point I talked about you know the total dollars going out the door as a percentage of earned premium were going up, going into 2017 and then started coming down in 2017. Which again, we believe is an indication that the things we've done using modeling to amplify you know what has always been a hyper-focused claims process is going to result in tail risk that would traditionally have been paid in the future being eliminated. But it's too early for those things to be given real credibility in traditional actuarial methods. And so we are obviously in a regulated business. And we made the decision to go to the select point but some of these things that we're sharing with you on the call are the things that gave us comfort that there wasn't an issue during the year but the reserved analysis came back as it did and again we took these extra steps so that going forward we can give very specific detail about how many claims from 2015 and prior we closed in a given quarter, how much we paid for those compared to what we thought and how much remaining IBNR is left for those accident years. And again, if we are paying the same or less than we thought, it will definitely be less than the reserves that we're putting up at year end. And then for the newer claims, we have these predictive model based case reserves put up at 45 days. We'll also communicate how many of those claims we've closed and how those claim closures have compared to what the models said it would be. And again, so far the claims we've closed that have been modeled in aggregate have closed for last and obviously that's something we want to continue to validate and we'll disclose. So going forward, having done that file reserve which we again were able to do this year because the claims were old enough and had enough information to make it credible and also now having the majority of our claims being having had cases set with the model will enable us to provide that additional clarification. And we are going to increase what we do internally in terms of our own internal actuarial staff which have really been focusing on pricing work to make sure that we're doing some regular, actual to expected testing and not relying entirely on outside actuarial in terms of the information that we disclose going forward. ...................................................................................................................................................................................................................................................... Q Okay. And then you mentioned that you are – you've moved your – the company's reserve pick to the outside actuarial select pick. And I guess the question is, where was it a year ago kind of year-end 2016 relative to that level? And historically looking back at prior accident years, how has that pick compared to what has ultimately developed on years that you can kind of consider fully baked at this point? ...................................................................................................................................................................................................................................................... A

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 10 Copyright © 2001-2018 FactSet CallStreet, LLC Yeah. I mean, if you look at two slides, really 5 obviously shows the amount of strengthening by accident year and you can see it's really concentrated in 2013 and 2015, you know, 2016 and 2017 there was some strengthening in certain lines but some redundancy in others. But to get specific with your loss ratio question on slide 12, the loss ratios we show here for our business which are the top three charts are fully developed loss and loss adjustment expense ratio, so it includes both ALAE and ULAE. And this includes moving to the select point from our outside actuaries in each of these accident years. And so if you look at Global Liberty exclude – Global Liberty is running about a 67% for both years; Global Liberty was about $50 million worth of our business in both years. For the ASI Pool Companies excluding Michigan, you can see that the fully developed loss and LAE ratio for 2016 is 59.5%, 2017 is 58%. If you weighted average the Global Liberty and ASI excluding Michigan, it gets you to about a 61% to 62% for those two years and we were carrying about a 60% in 2017 and I think about 61% in 2016. So again the more – the two most recent accident years So again the more – the two most accident years based on the same people doing the same work that determine the older years needed strengthening really come in, in line with what our expectation was. Obviously if you go back to those older years, the loss ratios are much higher than we had expected. And that is very consistent with what I think we're seeing across the industry as well. ...................................................................................................................................................................................................................................................... Q Okay. And then if you see at the outside actuary put now, would you have been at it as well a year ago and just the outside analysis and your internal analysis indicated that just reserves need to rise or were you below the outside actuarial point a year ago and so you kind of got the double impact of both the new analysis suggesting there need to be reserve additions as well as you as management selecting a more conservative point relative to outside actuarial? ...................................................................................................................................................................................................................................................... A At year end 2016, we booked two our outside actuaries select point. In fact on a net basis, we were a few million dollars above it. So it was not the case that we took an aggressive position last year and had to make up for it this year. We went to their select point last year. Last year, we did identify the issues that really did change in Michigan in 2016. And that was really our view and the conclusion that that was the need for strengthening, and again throughout 2017, we didn't see average paid severities remission continue to deteriorate. They continue to be high, six times our normal average severity. But the issue here wasn't that something dramatic happened in terms of paid claim activity, it was just a readjustment of the select point at year-end. So last year, we were at the select point, this year we're at the select point. But as I mentioned, you know, we did additional things this year in terms of the claim, review, and you are putting in place some of these other tools to make sure that we can not only confirm that what was done at year-end 2017 holds, but also start to more effectively show the positives that we think are going to begin showing up more clearly in the data, which is based on both on the intent of all these analytics tools we've implemented as well as is internal processes coupled with again some of the data points that you see on for example slide 11 where we are really seeing fewer claims based on more insured cars that we insured at higher average premiums and less total dollars going out the door relative to premiums earned. ...................................................................................................................................................................................................................................................... Q

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 11 Copyright © 2001-2018 FactSet CallStreet, LLC Okay. And one more question just shifting away from the reserves themselves into kind of a capital. You mentioned about subject to kind of rating agency actions. Can you walk, I mean obviously book value to pretty good hit here, capital of the company and you expect more strong growth. So can you walk us through kind of what levers you plan on pulling to kind of get everything back in line you're running at pretty full operating leverage. I know you have room on your quota share treaties. Can you just walk us through that kind of the shifting parts that you expect to have to move. And then if you think that'll be sufficient to at least on the year assessment of kind of the car and other measures sustain the ratings you currently have? ...................................................................................................................................................................................................................................................... A Sure. It's an important question. So obviously our customers are not particular in best rating sensitive. There are certain jurisdictions where there are rating requirements which is why we have the fronting arrangement that we've talked about before. And so today about $20 million to $25 million of the premium we expect to ride is going to be via that program already. And in terms of the- quarter share piece, with surplus in the, call it mid-80's, we would be gross written premium to surpluses like 3, 3.5: to 1, writing above $300 million, we do have the flexibility under our existing quota share arrangement to increase our quota share to as much as 50%. We did confidentially talk with our quota share partner. Once we became aware of the outcome here earlier this week, they have reconfirmed their commitment. The cost of that program is going to remain unchanged and they are prepared to support us up to 50% quota share. When we've modeled this out, it looks like we will probably increase quota share to around 30% or 35%. That is not a final number. It's something that we still need to tighten up, but we have the flexibility we have the support. And again we don't think even in the event of a downgrade that we would have an issue with customers. The market is pretty hard right now and appears to be hardening further both in commercial auto and in our niche. But obviously, we want to make sure that we can maintain a good track record in terms of our rating, particularly thinking ahead to 2 years, 3 years, 4 years, however, many years down the road when the market might soften and new competitors with higher ratings might come in. So, in the short run, the rating is not a super significant issue. The operating leverage is and again we have the support without having to even maximize the use of it to enable us to have net written premium below 2:1, which is kind of the guardrail. We've always said we keep in mind we're very comfortable with that we hope and best to be comfortable with that. And in any event, having the support of a strong reinsurance partner is obviously valuable to us and it has a cost to it, but it's an attractive cost of capital. it's an attractive cost of capital. ...................................................................................................................................................................................................................................................... Q Okay. And just one last one, can you just remind me where the quarter share sat currently as I know moved I just cannot remember that where it landed? ...................................................................................................................................................................................................................................................... A Sure. So in 2017 for global liberty it was 25% and for the remaining three companies it was 5%. ......................................................................................................................................................................................................................................................

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 12 Copyright © 2001-2018 FactSet CallStreet, LLC Q Great, all right. Thank you very much. ...................................................................................................................................................................................................................................................... A Great. Thanks for the questions. ...................................................................................................................................................................................................................................................... Operator: Thank you. [Operator Instructions] Our next question comes from the line of Bill Dezellem with Tieton Capital Management. Please proceed. ...................................................................................................................................................................................................................................................... William Joseph Dezellem Chief Investment Officer & President, Tieton Capital Management LLC Q Thank you. A group of questions, first of all, would you please repeat something that you said I believe it was in your opening remarks that gave you comfort in your reserves up until you receive the actuarial report please? ...................................................................................................................................................................................................................................................... A Sure. So when we began using the predictive analytics [indiscernible] our claims, the goal was to identify potentially large claims earlier and route them so that they could be evaluated if they were legitimate or meritorious get them paid earlier which generally for liability claims results in the lowest realistic payment or if they were inflated or not meritorious, start defending them earlier with the goal of extinguishing them down the road for zero or less dollars. And so, as those modeled claims became the majority of our claims, we were closing them a lot faster and again that's what we show on slide 9 is the claims that were settled between 10,000 and 25,000 and 25,000 and 50,000 which is really the two target cohorts. We're definitely closing faster. But when you look at slide 10, you can see we were not writing bigger you look at slide 10 you can see we were not writing bigger checks on average to close those claims, and the case reserve that was set by the model was generally higher than the amounts we paid. So in aggregate, for claims that we paid that had been modeled, and had case reserves set by the model, we were paying a significant percentage of last when those claims were closed with or without payment. And so when we looked at our aggregate information, the fact that older claims were coming in and being paid for more than case and using up IBNR appeared to be being offset by the release of case reserves from the model, with those accelerated closures on the large claims. And so, until the year-end work was done, it was difficult to bifurcate or look in aggregate and say what is the reserve need for the future for those older claims, because again we did not see average paid severity on a calendar year basis go up significantly from the prior year, that was what we did see in 2016 in Michigan. We did not see it again in 2017. We also saw frequency coming down. So, we had less claims coming in per car. We have less claims in inventory in total, which you can see again on slide 11. And so we had less claims from a frequency basis. We weren't paying materially more on average for the claims we were paying, and the predictive model appeared to be, and we think was working, but because of the acceleration it had on the bigger claims, it led to a total dollars going out the door going up before they start coming down, which again is really that what we try to illustrate on that chart on the left side of 11. So, yes, we were paying more claims and we were doing that on purpose. It's the goal of using the model to target those big claims and less claims were coming in per car, which suggests that we've been more effectively underwriting over

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 13 Copyright © 2001-2018 FactSet CallStreet, LLC the last three years which again lines up with when we started using predictive analytics in underwriting and risk selection on top of the expertise we've always had. And so all of those things made us feel comfortable that the reserves set at 2016 were going to hold up at 2017. And when the work was done obviously you know it led to a different conclusion and as I said we believe there are going to be positive outcomes as a result of these things that are starting to show up and now relate to the majority of our claims, but the outside actuaries we work with didn't feel that there was support for that at year-end 2017, and we felt it was important to move to their select point. ...................................................................................................................................................................................................................................................... Q Thank you. That's helpful. And then how much of the 2015 and prior business I'm going to call it problems were due to underwriting that was done by Atlas versus underwriting that was done by acquisitions? ...................................................................................................................................................................................................................................................... A So I don't have a specific percentage to answer that question and I'm not sure that we could pinpoint that. For Global Liberty, we bought the company in 2015. So if you just think of it pragmatically, we owned it in March of 2015. We did make changes specifically to the non-New York business as we've talked about we really bought Global for their expertise in New York. And so we did either run off or transitioned non-New York states to other areas of Atlas that had had more experience in those states. And so the [indiscernible] (00:41:19) years for Global I'd say the "old" process related to business that was written probably through 2015, may be a little into 2016, given that in our business, if you make changes at the rate filings or even procedural changes it takes time to sort of earn through the book" For the ASI pool companies, again it's harder to pinpoint, we owned all three of the companies going from 2012 through current and so I would not suggest that that was somebody else's problem. If you want to call it a problem, but as we talked about back down and we reminded people now from a pricing perspective we were on the high end of market. We always priced ahead of ISO as a proxy for what kind of the industry rate was based on our actuarially determined reserves and our outside actuary determined pricing recommendations. We were sufficiently reserving in pricing based on the information available then with the benefit of hindsight, obviously there wasn't enough premium generated to cover the cost of the claims there. If you back out Michigan, it is enormously better as you can see, if you just look at the standalone Michigan, but even then 2012 through 2014 would not have satisfied our margin expectations given what we now know. ...................................................................................................................................................................................................................................................... Q That's helpful. So if I may make sure that I'm clear, the Global Liberty you probably could pass that off as their underwriting problem, but the rest of the business you really do have to own that basically? ...................................................................................................................................................................................................................................................... A Yeah and it's a practical matter. You know we own on all of it. I mean, we bought brought the company, we diligence the company, I don't think it's appropriate to say this was somebody else's fault. I do think it's fair to say, what we are seeing is not inconsistent with what's happening across commercial auto over the last two years. But we also believe that as a specialist, we should be better and we should be able to be more nimble and react, and I think if you look at the results in all the companies over the last three years, although, the things we've done especially using analytics were not a reaction to what we then thought it was a problem that we was just trying to

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 14 Copyright © 2001-2018 FactSet CallStreet, LLC amplify what we thought was already good or great. But it certainly has had a positive impact, the last three accident years have shown consistent improvement, and again what's starting to flow through the numbers, you know we think is definitely not only supporting that the most recent accident years in our view in terms of future guidance, but also I think it does demonstrate the efficacy of the things that we've done in the more recent years. And you know we could not have done those things without the data and the expertise that we acquired with all of our companies. But obviously, we don't like the fact that book value just lost a significant amount, because of having to strengthen the reserves for those older years. ...................................................................................................................................................................................................................................................... Q Right. Great, understood. Thank you. I appreciate the clarification. And one additional question, and I apologize for my ignorance when asking this, but to – you mentioned that you're not getting credit for settling the claims in an accelerated manner, your more recent larger claims. The question is, is that – I'll call it good behavior. Is that actually hurting your actuarial outcome, because the models are looking at the dollars going out of the door and red flags are going off as – as you're paying more the red flags are going off as you're paying more right now and not taking into account the fact that you're simply paying earlier. Can you talk through that and to what degree the actuarial model is hurt by that phenomenon and you probably get the concept go ahead and just talk through that if you'd know see if... ...................................................................................................................................................................................................................................................... A Yeah. No. I guess it's very insightful question. I mean we do believe that that's happening. You know again if you look at the chart on the left on side 11 a very simple way to think about the actuarial work and I'm oversimplifying it is essentially is regression analysis where you're predicting the future based on the past. And so anytime there's a change good or bad, it can be challenging to identify. And there are different methods and different assumptions actuaries can make. But if you simply look at that chart on the left and think about the average trend you might draw as a line through you know starting about halfway through following that blue line, right, we would have an increasing slope. During 2017, you start to see it coming off, but the reserve indications we got in the strengthening we took half to assume that that line continues to go up not that it will continue to go down. We believe that the fact that it inflected and came down is exactly what we'd expect, because we paid the claims we're going to pay any way earlier and we paid them hopefully for less. And the ones we didn't pay that are still in inventory, we are defending potentially earlier, hopefully more effectively. And when those get paid down the road, they're not going to be paid for the big amounts that older claims typically got paid for. But that is ultimately the unresolved or unidentified question right now and so operationally we know that was the objective and we believe there's early indications in the data that it's happening, but we do not believe that that was fully contemplated or even significantly contemplated in some of the reserving work based on the kind of constraints of what you'd call generally accepted actuarial methods. So we have more confidence especially for those older claims, which again those are the accident years that, that our actuaries are telling us needed strengthening in our own reserve analysis. We've got people with a lot of experience who are very good at what they do in the claims area and we're very comfortable that, that file by file analysis is a good representation of ultimate liabilities for those claims. Again we'll track them closely. We are going to be articulating publicly every quarter how they're shaping up. And in the end once they're done, they're gone, right. As of June 1st, 2016 every liability claim, pip claim that we got for the SI pulled companies was modelled. We're transitioning global to that platform as well and we have a lot of

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 15 Copyright © 2001-2018 FactSet CallStreet, LLC confidence going forward that, that model based approach is going to be more accurate than having to rely on more traditional methods, especially given all the changes we've made in the business. But again at this point, it wasn't enough to try to reserve less than the select point and the appropriately conservative thing was to go to this point and try to share the facts with our stakeholders as transparently as we can. And again the good news is that the same people doing the same work with the same assumptions also looked at 2016 and 2017 and obviously came to a much more positive conclusion in terms of what we thought all along. ...................................................................................................................................................................................................................................................... Q So we put some words in your mouth that you did not say, but I want to make sure that, that maybe I'm understanding the concept here which is the actuarial report is really based off of statistical analysis as opposed to a lot of forward thinking, and that forward look is taking a rearview mirror and applying it to the future. Is that really the correct essence of this? ...................................................................................................................................................................................................................................................... A I think so. ...................................................................................................................................................................................................................................................... Q And that's helpful. And then the next question is, if you are correct and you are paying claims more quickly and therefore the future claims there's less of a bucket to pay. When do we start to see that show through from an actuarial perspective? ...................................................................................................................................................................................................................................................... A Well, you know, it's hard to answer that with precision. I mean if it continues to go forward with a clear trend. So let's say we stick with that chart on slide 11, that downward trend continues throughout 2018, I think it would be very hard to ignore especially if we continue to see claimed frequency flat or declining and claim inventories continuing to decline. But you know – that is a difficult question to answer with any amount of precision. But what I will tell you is we are going to be doing more work on an interim basis using our own internal resources and potentially also with other outside resources to make sure that we have as many different views not just optimistic, but also critical views just to continue to recheck these conclusions as we go through the year versus wait until the year-end for review to have another sort of bite at the apple. ...................................................................................................................................................................................................................................................... Q Great. Thank you for the time and taking all my questions. ...................................................................................................................................................................................................................................................... A No, I appreciate it. Thanks, Bill.

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 16 Copyright © 2001-2018 FactSet CallStreet, LLC Operator: Thank you. Our next question comes from the line of Robert Farnam with Boenning & Scattergood Inc. Please proceed. ...................................................................................................................................................................................................................................................... Robert Farnam Analyst, Boenning & Scattergood, Inc. Q Hey there. Good evening. Please proceed. ...................................................................................................................................................................................................................................................... Q Hey there. Good evening. I guess one of the questions I have was I understand you see the improvements and why you wouldn't pick below the external actuaries select point, if you're going to have such a disasters quarter, did you ever consider booking above the selected point? ...................................................................................................................................................................................................................................................... A So it's an interesting question. I mean I guess the bottom line is we've always tried to get it right. And we also have to work within the constraints kind of the rules of the game and so being at the select point, we think is appropriate based on the rules of the game. And I think going higher especially given that we have a lot of internal analysis that suggests otherwise would really just not be appropriate. I mean there are two ways to think about a reserve strengthening. On the one hand, it's a negative obviously for book value if it turns out to be correct and also for income in the short-term if it – but then also from a tax perspective right if we were to over strengthen especially if our own records suggest that that's not appropriate or we don't think it's probable now we've overstated for tax and particularly given that 2017 will have a different and higher tax rate than 2018 and presumably forward. I'm not sure that that would have been a smart move for us either. But the bottom line is we try to get it right and also play within the paint with kind of color within the lines based on the rules of our regulated industry and that's what we think the decision we made ultimately did. ...................................................................................................................................................................................................................................................... Q Did you ever consider getting maybe a second opinion as well. I mean given that this charge is going to be so big, did you reach out to maybe another third party another third party to maybe get an analysis done? ...................................................................................................................................................................................................................................................... A So we did do work with other outside experts many of actuaries not necessarily reserving actuaries to help us support what we believe was happening in the business. But I think anybody using a traditional actuarial approach to assess reserves would probably see similar things in the sense that, you do have old claims that are coming in and being paid for more than the reserves contemplated a year ago. And we do have these moving parts where big claims were being paid earlier and early evidence that the claims that are remaining are going to cost less than they used to. But I don't think that somebody coming and using a traditional reserving method is necessarily going to come in with a materially different number. But again I think

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 17 Copyright © 2001-2018 FactSet CallStreet, LLC one of the things we want to do is think about the fact that as we go forward not only do we want to continue to reaffirm the things that we believe are correct and share those with investors and other stakeholders, but we are going to bring in incremental opinions. I mean all these things cost money and so we think about what is a good use of capital in the short run, when there's only so much additional data. But obviously we don't like that we had strengthening year ago, we certainly don't like that we had it twice. We are going to do more to make sure that we have as many different perspectives and as much support for what we think are ultimately going to be these emerging positives, as we go into our year end work next year. And again we're certainly going to provide incremental information that that we weren't able to do last year this year. Now that we've done this file review for those older claims and now that the majority of our liability claims have been modeled and we think can – can serve as a as a point of reference as well. ...................................................................................................................................................................................................................................................... Q Okay. And just to verify your comments after last year's reserve charge, you didn't do a file by file analysis, because the – because the claims were not ripe enough to do that. Is that am I – did I hear that right? ...................................................................................................................................................................................................................................................... A Well, it was it was two things. It was that. Historically, before we started using predictive analytics and I think this is consistent with most commercial auto writers. A claim generally – a liability claim you know the kind that stick around for more than a year, generally had to be about a year old in order for it to really be estimated. Some claims could come in the door and you know immediately they're going to be a limits loss. But you know generally speaking, it took about a year for adjusters in the traditional way to get to a good estimate of ultimate loss. And so at the end of this year, at the end of 2017, every claim for 2015 and prior was at least a year old. When we look at our reported frequency, almost all of the claims for given accident year are reported within 12 months following the end of that year. So in other words, a claim for accident year 2015 is probably going to be reported with almost total certainty by the end of calendar year 2016. And so as a result, by the end of 2017, every single one of those claims that's still open is at least a year old. So part of it was you need to have enough information for the human beings to handicap that. And again that's part of the power. The modeling is it can handicap it earlier. And again that's a big part of why the current process is so valuable. But that was the – that's the first piece that you just need enough information for that kind of exercise to be useful. The second thing was, last year when we did the reserve strengthening, it was very clearly driven by Michigan and changes in paid claim activity in 2016 for that state. You know we literally saw pace severities on a calendar year basis double, even more than double for PIP. We saw an enormous increase in Attorney Represented claims. And so all of those things really pointed to Michigan was the problem last year. And so we focused a lot on Michigan, but we did not have a reason to think the rest of the book was going to have you know additional development. This year obviously, as we look at the year-end work and we look at where the reserves are strengthened, it's not line specific, it's not state specific. You know, our paid triangles don't show a particular hotspot anywhere. It is sort of an across the board you know the groundswell for those years and that's the actuarial estimate for what is necessary to pay all the rest of those claims. So you know that surprised us. It seemed higher than it made sense. And so that was we're saying, you know what; let's roll-up our sleeves and look at every single one of those claims, so that we can have our own view as to what those claims are going to cost and we feel very comfortable, that's a good number. And again that number is less than what we strengthened the reserves to.

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 18 Copyright © 2001-2018 FactSet CallStreet, LLC So as we look at those payments every quarter, if we're paying consistent or less than what we expect based on the audit on the internal power review that will be a very good fact because those amounts are actually in aggregate less than the reserves we're strengthening to. And if at the same time, the claims that we're paying that we're modeled, continue to be paid for less than the model set, that's also going to be a good fact because the reserves we're strengthening are also higher in aggregate than those case reserves. So you know that will be the process going forward. And you know, hopefully that helps to distinguish why it made sense to we did this year and why we didn't do it last year. You know, it really was, it was reacting to a different set of circumstances. You know, one that last two to three sort of state specific, whereas this year it's really just a broader, you know like I said in response to the earlier question, you know there's no sort of silver bullet, it was a particular thing, it was you know overall perceived increase. You know some of which we think is going to be real, but you know we think that we've really ring fenced it with this file audit and you know like I said we're going to make sure to provide transparency every quarter. So you know stakeholders don't have to wait a full-year to see how it's turning out. And obviously, we're going to look at it even more closely than that. ...................................................................................................................................................................................................................................................... Q All right, okay. And just one other on A.M. Best. So, typically A.M. Best allows companies that have pretty stable earnings to operate at higher leverage. If there's a chance that A.M. Best as you know, we don't want you're operating at two times book, two times surplus about one and half, how much does that quota share do for you if you do go up to the 50% and what would you do if it's still kind of short of where the investor wants you to be? ...................................................................................................................................................................................................................................................... A Well, if -- to answer the first part of the question, the 30% to 35% range I suggested would take us down to about 1.8 times. If we went all the way up to 50% on the quota share, we'd be a 1.35 times net written premium to surplus, so that'd be a pretty low operating leverage. ...................................................................................................................................................................................................................................................... Q Yeah. That's what I wanted. I want to be kind of get the sensitivity to the changes in the quota share? Okay. That's all I guess for now. Thanks. ...................................................................................................................................................................................................................................................... A Thanks for the questions, Matt. ...................................................................................................................................................................................................................................................... Operator: Thank you. Our next question comes from the line of Samir Khare with Capital Returns Management. Please proceed. ...................................................................................................................................................................................................................................................... A Samir? ......................................................................................................................................................................................................................................................

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 19 Copyright © 2001-2018 FactSet CallStreet, LLC Operator: Mr. Khare, your line is live in conference. Please proceed with your question. Samir Khare with Capital Returns Management. Please proceed with your questions sir. Thank you. Sir, is your line perhaps on mute. ...................................................................................................................................................................................................................................................... Operator: Thank you. Sir, is your line perhaps on mute? Thank you. Our next question will come from the line of Cliff Gallant with the Philadelphia Financial. Please proceed. ...................................................................................................................................................................................................................................................... Cliff Gallant Q Thank you. One thing I didn't understand is if the outside actuaries at the end of the day are going to be ones who set the loss picked at the end of the year, why aren't they more involved in the process the 3, 6 and 9 month reporting periods? ...................................................................................................................................................................................................................................................... A So historically, we have had them do full yearend reviews. We have had them do interim pricing work during generally the first half of the year. And then that pricing work informs our average rate change. And then typically would have some preliminary analysis done after 9:30 and then the full year review, part of it was for the size of the company we had in the amount of incremental data that would be accumulated over the course of the year, leading up until this year there really had not been any significant reserve changes. And so there really wasn't a sense that they're needed to be more interim work done and as a practical matter for the size of the company it may not have been that useful. Last year because the strengthening was related to a specific area, not kind of a broad issue, again the focus was more let's make sure we're keeping our eye on that state and as we said it was not developing worse than it had really the year before in terms of average paid. This year given that we are looking at what we think is more of an industry issue related to 15 and prior, we're going to do that. I'm sure at this point it sounds a little bit like kind of is it too little too late, but the reality of it is last year we weren't reacting to a situation that we now are seeing. We have a lot of reasons to believe that our business was going to behave better than commercial auto in terms of limit profile, in terms of the vehicle weights, a number of other things we've talked about in the past and we continue to think that ultimately it is better than average for commercial auto, but I'm sure you've been looking at commercial auto result as well. Overall, the industry has had a very tough last few years in terms of development going back to the same exiting years. And obviously, some of those issues that were affecting the broader industry, clearly impacted our book more than we thought and more than our actuaries thought a year ago. But now that it has come to light, we are going to do more work and to be clear, we really have a tail to cities here. We have the old stuff that's running off that was not modeled and we have the new stuff that is increasingly the majority of our business that was modeled. The old stuff is performing worse than expected, the modeled stuff thus far is performing better than expected. And so part of our strategy is let's quarantine the old stuff which again we think we've done through, the file audit and the additional things I've talked about and let's make sure that we can continue to revalidate the good stuff, because as a practical matter, we don't think that got a significant amount of any credit in terms of the year-end reserving. We understand why, there is a limited amount of data because modeling claims in particular, really started taking effect only mid-year 2016. But again, it is now the majority of our business and so for what we do in

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 20 Copyright © 2001-2018 FactSet CallStreet, LLC 2018 and forward, it's going to produce a lot more data points and I think is going to be a lot more credible going in the year end. But we want to have a lot more information between now and then to make sure that we don't again find ourselves surprised by an outcome that that sort of -- was inconsistent with what we were expecting. ...................................................................................................................................................................................................................................................... Q The one thing, one of the things you were talking about in the Q&A was in terms of looking at the this year, it sounded like what you said, and I just want to clarify that the total paid losses were coming in as you expected, what you didn't appreciate was that – that a high percentage of it was going – was coming from the old accident years. Is that correct? ...................................................................................................................................................................................................................................................... A Well, I think in terms of – how much of the older accident years using up IBNR exceeded what the newer claims were doing to help IBNR. Because if you think about what was happening at 45 days, we are putting up case reserves for claims that for those claims under the model of the environment, those claims that had been paid were being paid for less than case. So, on a incurred basis, that the reserving process itself significantly replaced IBNR with case, and then the older claims that were being paid that had the lower factor reserves were actually legitimately using up IBNR on a paid basis. And in aggregate there was sort of an offsetting effect and our expectation was as the run-off piece of that non-modeled claims got less and less, it's deterioration on IBNR would be more than offset by the benefit from you know what was going on the modeled claims, but based on the year end work, the conclusion was there was not enough benefit in the new stuff. In fact there may even be a reserving review that there it may have suggested the [ph] LDFT to be bigger than we think they are. And at the same time the assessment of what's left in that old inventory actuarially is just bigger than our own team has assessed when they've looked at those – those individual cases by file. So you know it is a different of opinion but it's also a different approach, it's a file-by-file review versus an actuarial assessment of 2015 and prior in the aggregate. And again the only thing that will resolve that difference of opinion is time and you know we will be communicating how those are you know how those claims are working through the system on a quarter-by-quarter basis because really it's the only way that we can ultimately demonstrate what the final outcome is going to be, you know you've got one essentially one estimate that's based on math, and you've got another estimate that's based on a granular look at every single file. But as with any claim in the end the final outcome is the thing that matters. And so you know, we want to just make sure we both have a very strong control environment internally, particularly for those older claims which is where the problem seems to be and then you know be able to articulate that with great transparency to our stakeholders, you know shareholders and others as we go forward and again you know make sure we've got the support to get more benefit from or more credit for the benefits that we are actually seeing emerging in those neurasthenic years. But don't really think are reflected in the – in the reserving decision. ...................................................................................................................................................................................................................................................... Q One last question. There has been a shocking amount of value destruction at the company in the last – over the last two years. At what point do you and the board consider strategic alternatives for the company? ......................................................................................................................................................................................................................................................

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 21 Copyright © 2001-2018 FactSet CallStreet, LLC A Well I think the most important thing from my perspective is you know we've a business that has a strong amount of value, in terms of you know expertise data, the use of analytics that you know ironically today may actually be creating a perceived issue that is hurting, you know the current book value, but the reality of it is, the value drivers in this business I think are stronger today than they've ever been Obviously, we need to have a balance sheet in order to continue to grow our business and we are in a hard market and we think it could be hardening based on all the results and all the other information that's been communicated by the commercial auto industry in the last – really the last year and a half, but – but including the last quarter. And so, we want to optimize the value of what we have in the environment we're in. And if, there was some kind of a partnership that would allow us to maximize that value, obviously you know the board would have an obligation and I think you know collectively the board and management would have an interest in considering that. You know I am certainly not suggesting we are going out and looking you know as I said we have good support from our quarter share partner. We've got the ability to write the amount of premium that we think will optimize return on equity in 2018 with that support, without needing to go out and find incremental capital. You know we've got the data, we've got you know unfortunately having you know gone through these challenges, we've got even more insight and information into to how to make sure that our future business is as profitable as possible. So, you know as we sit here, I don't think we need a partner to do something strategic or consistent with our current strategy. But that doesn't mean we wouldn't be open minded if there were something that was going to allow us to take those value drivers and make them even more valuable you know for our business and for our stakeholders. ...................................................................................................................................................................................................................................................... Q Thank you. ...................................................................................................................................................................................................................................................... A Thanks for questions, Cliff. ...................................................................................................................................................................................................................................................... Operator: Thank you. Our next question comes from the line of Brian Hollenden with Sidoti Please proceed. ...................................................................................................................................................................................................................................................... Brian Hollenden Analyst, Sidoti & Co. LLC Q Yeah. Sorry, if I missed this, but with the reserve strengthening was added on more concentrated neither taxi, limo, or para-transit or was it broad based? ...................................................................................................................................................................................................................................................... A It was pretty broad based. On slide 5, the slide on the left which is the ASI Pool companies, we show this strengthening by accident year by line of business and also we segregated out Michigan. And so what you can see is taxi probably has the most strengthening across all the years. Obviously that's the aspect of our business that's been shrinking.

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 22 Copyright © 2001-2018 FactSet CallStreet, LLC And the next largest is of course Michigan and then it kind of varies by accident year. So – it wasn't other than those two things, it wasn't any one thing consistently across the accident years and again our internal process in terms of pricing and modeling is much more granular than this. And so notwithstanding reserve strengthening, you know, we are always looking at every line of business in every state and evaluating what we need to do to optimize the return on deployed capital from an underwriting perspective. And we are seeing continuing shifting towards more predictably profitable account which is what we show on the right side of slide 4 and so obviously we are going to revisit all of our pricing work with these new reserve levels. We are going to also consider other factors in determining the best way to deploy the capital we have to optimize the return on that. But as I said earlier, it wasn't a sort of single line of business that suddenly went from good to bad, it was really kind of a mixed bag in terms of what we received in terms of the reserve study. ...................................................................................................................................................................................................................................................... Q Yeah, one last one for me. Just what gives you confidence that the reserves won't need to be strengthened again next year? ...................................................................................................................................................................................................................................................... A Well, I think it's probably three things, and I've tried to articulate them. But to be clear, the first is the strengthening here is really focused on these older years and those claims are now mature enough that we really feel we can handicap them and that's what our file review is intended to do. We know the conclusion of that review is that the ultimate liability for those claims should be less than the reserves we now carry for those accident years with this strengthening, so that's the first piece. The second is the case reserve set by the predictive model, which is claim specific has been behaving as expected. It has shown redundancy when we've compared actual page versus case and we've had some additional outside experts to take a look to validate that, that does appear to actually be a long-term phenomenon not just to short-term. And then the final piece is, is that the same approach that concluded that there needed to be strengthening in 2015 and prior, also concluded that 2016 and 2017 were more or less consistent with what we thought. And again, that lines up with the timeline of the things we did in our business to be more precise and effective with risk selection, to take rate in a hardening market. And it also lines up with the reported frequency in our existing claim inventories. So we're not seeing claim inventories building on a year-over-year basis despite the fact that our of book business grew, we're seeing reported claim frequency down and we're seeing relative stability in terms of average page kind of on our development triangles in aggregate. So all of those things together give us the level of confidence that we have, and on top of that again, we do not believe that some of these things that are going to be emerging positives could really be contemplated in some of these traditional methods, not because we didn't identify or articulate that we think they're happening, but because it just doesn't sort of fit into the rules as it were. ...................................................................................................................................................................................................................................................... Q Thank you. ......................................................................................................................................................................................................................................................

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 23 Copyright © 2001-2018 FactSet CallStreet, LLC A Thanks for the questions. ...................................................................................................................................................................................................................................................... Operator: Thank you. Our next question comes from line of Sam Hoffman with Lincoln Square. Please proceed. ...................................................................................................................................................................................................................................................... Q Scott, hi. Can you break down your guidance submitted the target combined ratio in terms of – is that the traditional 27% expense ratio, a 58% loss ratio or is it 60% to 62% loss ratio that you're booking for 2017 with a lower expense ratio because of the reinsurance? ...................................................................................................................................................................................................................................................... A So we would expect and again these are pro forma estimates since we have not released our audited numbers. But looking at those fully developed loss in LAE ratios that again are after this strengthening on a blended basis you would expect the fully developed loss in LAE for 2017 accident year to come in around 60% to 62%. And that's assuming there is no incremental positives from the things we've talked about. We are continuing to take rate and we're continuing to see the market harden. So we are going to price to a better number, but the mid-80s loss ratio essentially assumes around 60% fully developed loss ratio for the full year 2018 and an expense ratio in line with that 26% to 27% number that we were at throughout the year in 2017. It will be slightly lower the expense ratio if we are using more quota share, because as you identify there would be a ceiling commission benefit, but all in, that's really the basis for it and to set expectations, we will come out in the first quarter and be more conservative and we will share exactly what's happened with claims that were settled in the first quarter, related to both our claim file review, as well as what our predictive model said for the newer accident years and what we actually paid. If we see those things continuing to trend in a positive direction, that will give us the confidence that we should move the loss ratio down to be more consistent with what it was last year, plus potentially taking benefit for the additional rate that we expect to put into the market this year. So I do want to be very clear about what we're trying to guide to – it would not be appropriate, I think for us to come out of the gate here and say, you know what, the last 2015 and prior were bad, the last two years look good. And so we're going to book a number in the first quarter that's even better. That may well be true, but we are going to build a bit of a margin of safety and certainly in the first part of the year. And then, as we are supporting these conclusions, that we believe will lead to a better outcome obviously it will makes sense for us to then ratchet down the result, but full year, I guess that our expectation is that that mid-80's combined ratio is reasonable. We've got the support from reinsurance in terms of the business and we're comfortable that that's a reasonable sort of minimum expectation. ...................................................................................................................................................................................................................................................... Q Okay. So right now, you're at the 60% to 62% level, you'll start somewhere within that range and then hopefully by the end of the year, you will be below that range?

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 24 Copyright © 2001-2018 FactSet CallStreet, LLC A Yeah. Yeah, I mean the calendar year result, the GAAP result is going to be really a blend between the accident year 2016 and accident year 2017 fully developed. And then, you know obviously, it'll you know we hope that it'll turn down going forward for the reasons I touched on. ...................................................................................................................................................................................................................................................... Q Okay. So my other question is, can you talk about loss trends generally in your business, because you know I've heard from some other competitors that you know that the tax season when those you know loss trends hasn't has been accelerating you know particularly in the fourth quarter and into 2018? I'm just curious, what you're seeing and if you're seeing things and what the causes of that car? ...................................................................................................................................................................................................................................................... A Are they setting getting increased frequency or severity or a combination of both? ...................................................................................................................................................................................................................................................... Q I think it's a combination of both? ...................................................................................................................................................................................................................................................... A Well, we're certainly not seeing increased frequency again, you know we put the numbers out on slide 11. And so, whether that's because our risk selection is better because we're using the predictive analytics and as far as I know nobody else in this space is, or just better judgment, I don't know. But we are definitely not seeing frequency increase as evidenced by the actual reported recorded claims per car year and the open inventory. Now some of the inventory is going down, because we're closing those bigger faster also, but the bottom line as is the reported car years would be an indicator, and we've looked also at our report date versus claim date and you know we are not seeing any [ph] elongation of reporting times on these liability claims. So you know it really is better frequency relative to insured cars, now that's for our book, right. I can't speak to what the industry trend might be, because again, we're using modeling in terms of risk selection and others are. In terms of the severity, you know again, we look very closely at pace paid severity triangles and we are not seeing a big increase in the annual severity triangles. You know we're still obviously going through more granularly quarterly information and by state by segment. And so I so I can't really speak at this point to what we might find in a more granular basis, but overall like I said on a year-over-year basis it's not a big paced severity jump that's leading to our view here. But clearly for those older accident years commercial auto ended up being worse than everybody thought. And unfortunately like I said we did not – we were not sort of insulated from it as we thought we would be based on the profile of our business. ...................................................................................................................................................................................................................................................... Q

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 25 Copyright © 2001-2018 FactSet CallStreet, LLC Okay. And my last question is, on the period 2015 and 2016 on [indiscernible] business. When do you think you'll have a complete knowledge of what the ultimate loss will be? ...................................................................................................................................................................................................................................................... A Well we've, we've done - ...................................................................................................................................................................................................................................................... Q Is there like a statute of limitations or a certain time when like three years or four years when you pretty much have a virtual 100% certainty? ...................................................................................................................................................................................................................................................... A Well the bottom line is that there is really no statute in terms of the adjudication of the [indiscernible] (01:21:43). I don't think Joe is there, I mean there's statutes were reporting in some places. ...................................................................................................................................................................................................................................................... A Right. And statute Sam vary by venue. So different states have different statutes of limitations, but if we've got the claim on our books and it's in the older part of that 2015 and prior, chances are good that it's in litigation. So that, that preserves the statute of limitations. But I think your question is really when are we going to be certain about how good or not good our [indiscernible] (01:22:17) it is and that's really like Scott has said just will happen over time. ...................................................................................................................................................................................................................................................... A Yeah and we will report that out every quarter on earnings call and again I want to make sure there's absolute transparency to that. However it works out. But again, we've a lot of confidence in the people that did that work. ...................................................................................................................................................................................................................................................... Q I missed the piece about the rating agencies , have you spoken with them and have they given an opinion or schedule for the reaction? ...................................................................................................................................................................................................................................................... A We have confidentially spoken with them. They will issue a press release when they come to a decision about what their reaction is. And that's really all the information that we have at this point. But again it's fortunately in the short run the market hardening, our customers are not particularly rating agency sensitive. Some of our agents though do look at that as an important issue. And so, we do want to manage that dialogue. We want to make sure they have good information; they understand what we're going to do with quota share that we are able to get credit for the better result et cetera in the most recent year. So, we are very – we will have a very active dialogue with them, but at this point I really don't have more information on that.

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 26 Copyright © 2001-2018 FactSet CallStreet, LLC Q Okay. Thanks ...................................................................................................................................................................................................................................................... A Thanks, Samir. ...................................................................................................................................................................................................................................................... Operator: Thank you. Our next question comes from line of Paul Newsome with Sandler O'Neill. Please proceed. ...................................................................................................................................................................................................................................................... Jon Paul Newsome Analyst, Sandler O'Neill & Partners LP Q Thanks. Thanks for your patience. I just want to circle back here a little bit on some of these general topics that we've talked about. It sounds like because of the acceleration in the claims being in settlements that you've been doing internally, that is causing a little bit of a difference of opinion between the outside actuaries and perhaps you in terms of the conservativeness of the reserves. And I just want to make sure that's right. And that if things turn out the way you think they are and you should be able do better than the – the company should be able to do better than the reserves you're going to put up based upon the outside actuaries. Is that just a fair assessment or am I misreading it? ...................................................................................................................................................................................................................................................... A No. I think you're interpreting what we're saying correctly. And again our goal is to be able to show evidence or incremental evidence of that in a way that can be actuarially supported as quickly as we can. I mean the first key is continue to deliver the results we're seeing operationally, which are favorable and then make sure that that information can be utilized both because there's enough of it, right, which is part of it, it's just credibility, there are enough data. And that we are making sure that we're able to share it with whomever has to understand it in a way that it can be contemplated as we go forward. ...................................................................................................................................................................................................................................................... Q And then I also wanted to know that the $2 or better number that is put out there for potential earnings, is that number incorporating what you would have to do – you may do with the quota share reinsurance or with the quota share reinsurance have a material impact on the bottom line profitability if you end up increasing to the 35%-ish range you're talking about? ...................................................................................................................................................................................................................................................... A Yeah, the $2 assumes about a 30% to 35% utilization of quota share. If we ended up using more, it could have an impact. But our cost for the quota share is obviously only a fraction of our margin or portion of our margin. But we are comfortable that if we're using a level of quota share, that's in or around that 35% range, that is a reasonable estimate. And again, if we use 35%, we would get down to about 1.8 times operating leverage and that's well below the 2 times that we've always felt was a sort of a guardrail.

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 27 Copyright © 2001-2018 FactSet CallStreet, LLC Q Good, thank you. ...................................................................................................................................................................................................................................................... A Thanks for the question. ...................................................................................................................................................................................................................................................... Operator: Thank you. Our next question comes from the line of Ron Bobman with Capital Returns Management. Please proceed. ...................................................................................................................................................................................................................................................... Ronald David Bobman Analyst, Capital Returns Management LLC Q Hi, good evening. I got a few questions. First one is sort of I think simple. I'm looking at the pricing change chart. We've got each quarter of price movement versus the ISO rate changes. And I'm surprised just one thing jumped out at me that the magnitude of the change like for example you've got 20% rate increase it looks like in the second quarter of 2014 on average for your book and it drops dramatically just one or two quarters later and then again in the summer of 2016 same sort of thing. It's plus 20 and within two quarters, it's down to zero. I'm assuming you've got sort of bunches of business that are concentrated in certain quarters that can create that degree of volatility. Is that the case or is it something else driving it? ...................................................................................................................................................................................................................................................... A So the first thing I'd point out. These are sequential rate changes quarter-over-quarter. So going down to zero isn't a reversal. It's just not taking more. But those two sort of bulges I call it that you're highlighting were for two different reasons. Obviously the market began firming up in 2012 as you can see even before ISO set raise rates we were. And then, you may recall that in 2012 and 2013 is when we saw a lot of MGAs lose their markets. And so a significant amount of the customers in our niche were having to find new insurance. And so we pretty opportunistically took rate at the beginning of 2014 to make sure that we were getting as much rate as we could get given the fact that there was this big dislocation of MGA capacity in the market. And then, we kind of came back down to what was more or less on pace with the market. And then, in 2015, it was a different dynamic. So you know and I know you pay close attention to the insurance industry. You know at that point mid-2015 is when there were people saying you know what commercial auto is going to soften along with everybody else. We were not seeing that in our niche. We actually expected that there will be sort of a rate retrenchment. So we again increased our rates pretty significantly expecting that the market was in fact going to do the opposite of what people thought, it didn't initially and our hit ratio has dropped significantly in fact we talked a lot in the first half of 2016 about that. So there was a combination of taxi shrinking, but also our hit ratios have gone down, because we had taken kind of more rate than there was tolerance for. So we went down to stay flat until the market did rise, which it did. At which you obviously then start to see happening at the end of 2016 and into 2017 and of course now we all know that the market did retrenched and in fact the commercial auto rate increases for the industry in 2017, in fact exceeded what you know were achieved sequentially in 2015 and 2016. So those are really two examples of us being sort of opportunistic based on our observations about the market cycle in the competitive environment,

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 28 Copyright © 2001-2018 FactSet CallStreet, LLC more than anything else. And again for kind of two different reasons, but that was sort of the – that was the rationale there. ...................................................................................................................................................................................................................................................... Q Okay, thanks. And as a sort of further clarification on that question, so you made a point of saying that these are sequential rate increases. So obviously for the most part, the policies are annual policies. So if I look at any one quarter, if we could just take Q3 of 2016 the average rate increase was zero. The preceding quarter was like 10% I think I'm reading this, right. ...................................................................................................................................................................................................................................................... A Yes. ...................................................................................................................................................................................................................................................... Q But that insured who renewed his account his coverage sometime during the third quarter of 2016. What does this chart sort of tell me approximately the rate increases that he faced or didn't face Q3 of 2016 as compared to when he renewed Q3 of 2015? ...................................................................................................................................................................................................................................................... A So assuming it is the same driver with the same risk characteristics, and let's say he's average, he or she, and let's assume that we only wrote a single line of business in a single jurisdiction. What that would mean is that the renewal or the new premium, let's call it a renewal, in 2014 would have been a 10% increase or I'm sorry in this case it was 5% increase, whatever it was. Yeah, which quarter do you want to start with? ...................................................................................................................................................................................................................................................... Q Well I'm just looking at Q3 of 2016, because the chart went to zero. And like you said... ...................................................................................................................................................................................................................................................... A Right. ...................................................................................................................................................................................................................................................... Q ... if it's homogenous as compared to Q3 of 2015? ...................................................................................................................................................................................................................................................... A Yeah. Okay. So Q3 of 2015 would have been call it a 7% increase. Q3 of 2016 would have been neutral with the prior year. So across those two policy periods, it would be about a 3.5% total increasing rate.

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 29 Copyright © 2001-2018 FactSet CallStreet, LLC Q Okay. ...................................................................................................................................................................................................................................................... A But again, this is not all aggregated too. So this is all lines, all segments, anytime there's an argument. So it's – but if you could make it specific to a particular individual that's the way to interpret it. ...................................................................................................................................................................................................................................................... Q Okay, thanks. And I've got some questions. The operating efficiencies, sort of the accelerated and improved claims handling, the predictive analytics, the machine learning, what order of magnitude are you hoping these initiatives deliver as far as combined ratio improvement? And I know, we're not talking about 90 days, we're not talking about 12 months, I assume it's sort of a 36 months sort of timeframe. What sort of order magnitude do these initiatives have the potential to lower your combined volume? ...................................................................................................................................................................................................................................................... A So over the last two years for the pricing, we believe that we have achieved an approximate 4% improvement in risk, rate relative to risk by using the analytics and pricing, which is causing us to bias towards the better accounts, which is what you see in those two charts on slide 4. So over time that should be a 4% improvement. So, if we were pricing on average using a traditional approach to say a 60% loss ratio, over time that theoretically would result in those call it 56%. So that is solely better risk selection and pricing. Separately in claims, there are two things that we do not believe are currently reflected in our current results. The first is redundancy in case reserve. So I mentioned for claims that we have opened while we've been using modeling, we use the model except the case reserve and we've now closed. In total, we've paid anywhere between 30% and 50% less than the model said we were going to for those claims. Some of that, a good chunk of that are claims that were modeled, every claim that gets modeled gets a case reserve put on it and we closed 20% to 35% of our inbound claims without payment over time, so especially in the first year to two years. So some of that 30% to 50% is just the effective effect of CWOP, case reserve gets put up, claim gets close without payment, it gets released back into IBNR. So, that's the first piece. Now, that won't hold. We're not going to pay 30% or 50% or less over time. So, we did hire an outside firm from doing analysis to help us validate what it probably would be over time their initial assessment, and again it's on a relatively small amount of data, but a year-and-a-half worth of data suggests there's probably about 18% redundancy in aggregate on the case reserves set by the model. So that's the first piece. It's not real dollar savings year-over-year, it's just a potential reflection of what you might call embedded IBNR in our current model case reserves. The second thing which was the goal of using the modeling and claims is actual reduction in pay severity over time. And so we did work with Guidewire, predictive analytics who is our partner who we built the models with and

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 30 Copyright © 2001-2018 FactSet CallStreet, LLC they compared claims that were modeled when we built the tool. And so when we built the tool we built it on about 70% of the data and we set aside about 30% that was developed enough data that it could be scored. And then when the model was built, it was test against that 30% so we have a database of claims that were scored and then settled without the benefit of the model helping our claims department two years and route those claims. We then compared those claims, so claims was similar scores that were actually identified when the model was implemented and settled. And it suggests that as a result of using the model to [indiscernible] 01:35:37 and route claims we may be saving as much as 10% to 17% in ultimate claim cost. Again, that's based on about a year and a half worth of data. It's relatively early. I think you can't – we can't rely on it. I wouldn't suggest we're going to carve 13.5% out of our loss ratio because of it. But it is directionally very encouraging and it's certainly consistent with the goal of using the model in the first place. So that's – that's probably the most precise way I can answer the question. It's all positive and in the end we think all of these things are potentially sustainable, because we're not competing with other companies who are also using modeling, it's not a war of attrition and who can do it better. It literally is at least for now, we are competing in an environment where our competitors are putting a commodity out there and we're differentiating to the use of these tools, and combined with the expertise and specialization of the company in the first place. ...................................................................................................................................................................................................................................................... Q Right. So, sort of speaking on sort of on maybe a sensitive subject, but on the topic of sort of expertise. So, obviously sort of in the area of claims setting and the adequacy of claims setting and the monitoring of litigation and the setting of reserves you said obviously it's sort of far from showing your expertise. So, is there are staffing changes internally? You talked about obviously hiring the outside actuary and doing the full report, but are you making any staffing changes internally adding talent? ...................................................................................................................................................................................................................................................... A So you know obviously, we're always looking to add good talent. I don't believe that this is a result of somebody in our organization made a specific mistake. I mean if you look at where the increases are it's for accident-years 2015, 2014, 2013 those case reserves were set to a statistical factor based on the average incurred for claims of a specific type in a specific area at the end of the year and they were held constant. And the IBNR that was developed by our outside independent actuary was intended to cover the development on those claims as well as any future claims that came in. And that's a process that is not dissimilar to what most traditional smaller mid-sized insurance companies do. And so again, I'm not laying the blame at anyone's feet including our outside actuary but the fact is. Those reserves were based on the development of those claims at that time, nobody has a crystal ball, we'd like to think that this modeling helps us get closer to that. But the reality of it is, we are obviously taking this seriously but this isn't an issue where I can go back and say you know X person or X department made a mistake back then, you know, again we were pricing ahead of the industry, we were relying on the collective available information at the time. If you think about our growth rate, you know, we were not clearing the market so that there was no anecdotal reason to think that, you know, we were underpriced by half. I mean if you look at the developed loss ratios today that is effectively the conclusion. If that were true back then and we and the market knew it we would have grown 200% in one year.

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 31 Copyright © 2001-2018 FactSet CallStreet, LLC So, again I know that doesn't give comfort to the impact that this strengthening has in book value today, but I do think it puts context around, I don't think this is a human being made an error. And furthermore I would point to look at the last three years which again the same people who were extraordinarily critical if you want to describe it that way of 2015 and prior are the same ones who use the same process to look at 2016 and 2017 and obviously they're coming to a very different conclusion. Those results are the result of the work that were done by the people in this organization for the last three years to four years. And I think that – and it is clearly better than the industry. So I think that is the best representation of the expertise and the quality that we have. But we are as I said going to increase the amount of disclosure we can provide around these specific comparisons based on the stuff that I've already you know and I don't want to sound like a broken record, but we have done things differently this year to make sure that we can communicate not only that the reserves we currently have are holding up, but ultimately we hope to have some redundancy if these positive things turn out the way that we expect them to do that more regularly and with more rigor and of course we are continuing to look for good people to make sure that we can continue to build on the good things that you see in exiting year 2016 and 2017 that you see an accident year 2016 and 2017. But that's to continue to improve the organization not because I think we have to go back and address something that somebody specifically didn't do right three years ago or four years ago. ...................................................................................................................................................................................................................................................... Q Thanks. I hope you're right. Obviously the 2017 and 2016 years and the data are far greener and so it's just time will tell and hope for all our cases. You're right, I know a couple of questions. The Michigan PIP book, how many open claims remain and also with respect to sort of the global liberty problem action years, how many open claims remain and then what's the corresponding IBNR for each of those sub books please? ...................................................................................................................................................................................................................................................... A So if you look at total claims, the total inventory at global liberty at year end was about 2,000 claims. I don't have the number by accident here in front of me. And we have as we did in the ASI pool we did have the experience folks in New York look at all those older accident years claim by claim. And so again we have the same ability to compare the resolution of those claims against the expectation, it is going to tie more closely to the reserve strengthening we took at Global and so the majority of the strengthening was related to the other companies not global and then in terms of the ASI pool if you look at the total inventory of non-modeled claims at year end, there was about 1,600 of those Michigan was only about 150 including some newer claims in Michigan that were scored claims that came in from the beginning of 2016 to date. There's a total of about 400 open Michigan claims at yearend 2017, that number is down from about 800 or so at year-end 2016. So it is definitely declining sharply. We do not get from our actuary-specific buckets at that granular level. I mean we will allocate it proportionately based on open exposure, but again the comparison we're going to do is what – is what the outcomes are versus our claim file audits. And those file audits in aggregate are predicting a lower payment amount than the reserve strengthening here suggest. So if we are within our expectation, we will be below what the current strengthening suggests we need and that's what we'll be disclosing. ...................................................................................................................................................................................................................................................... Q

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 32 Copyright © 2001-2018 FactSet CallStreet, LLC Okay. Would you hazard a guess or and it is my last question, if we roll forward 12 months, the Michigan – the portion of the Michigan book which is the troubled cohort as opposed to the this cohort in the more recent book, if I look at the troubled cohort what would you expect that open claim count to be down to at the end of calendar 2018? ...................................................................................................................................................................................................................................................... A I mean do you have a 01:43:34 a guess on that one. ...................................................................................................................................................................................................................................................... Q Or am I asking a [indiscernible] question? ...................................................................................................................................................................................................................................................... A It dropped almost by half year-over-year that that should accelerate a little bit. ...................................................................................................................................................................................................................................................... Q Oh, really [indiscernible] not slow. ...................................................................................................................................................................................................................................................... A Well we're down to we have no – we've less than 1% of our book from a premium perspective as Michigan now. So it shows this the inbound claims will decelerate sharply. ...................................................................................................................................................................................................................................................... A Yeah. I mean we did see a slight drop last year the attorney representation for Michigan claims was in the mid 60% which was like more than three times our national average. At year-end 2017 it is gone down to like those low-50s, still high comparatively but it's trending a little bit cooler. And obviously as Joe pointed out and we illustrated in the last deck of – side of the deck, the inventory in Michigan dropped off dramatically in 2017 as we expected it would, based on all the things that we did to make that happen. And so the in-bound is definitely coming down plus, unemployment is much higher now than it was in Michigan and Detroit, back in sort of those accident years where we had a lot of exposure there. And in terms of who can make a claim against PIP, obviously employment and having insurance has a big bearing on how many PIP claims you get for every hundred accidents. ...................................................................................................................................................................................................................................................... Q Okay. Thank you. ......................................................................................................................................................................................................................................................

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 33 Copyright © 2001-2018 FactSet CallStreet, LLC A Thanks for the questions. ...................................................................................................................................................................................................................................................... Operator: Thank you. Our next question comes from the line of m Matt Carletti with JMP Securities. Please proceed. ...................................................................................................................................................................................................................................................... Q Hello. Can you hear me? ...................................................................................................................................................................................................................................................... A Hey, Matt. ...................................................................................................................................................................................................................................................... Q This is Carl actually on behalf of Matt. And then just one question. ...................................................................................................................................................................................................................................................... A Hi, Carl. ...................................................................................................................................................................................................................................................... Q Hi. And then the question is what impact will these results have on the management compensation and bonus pool for the year? ...................................................................................................................................................................................................................................................... A Well, as we've always said, the management compensation program from a cash perspective is based on return on equity. We will not have a positive return on equity for the year. And therefore there should not be executive bonuses for sure, based on the plan and the more significant incentive compensation that we have and was filed a couple of years ago via 8-K is a stock and option compensation plan which is tied specifically to ROE and its impact on book value. And so for that plan to fully vest which there's about two years left on it, book value would have to be up in the 20s, low-20s. and so obviously given this adjustment to book value we're not going to move in the right direction in terms of investing in that program. ...................................................................................................................................................................................................................................................... Q

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 34 Copyright © 2001-2018 FactSet CallStreet, LLC Okay. Thank you. ...................................................................................................................................................................................................................................................... A Thank you [indiscernible] . ...................................................................................................................................................................................................................................................... Operator: Thank you. Our next question comes from line of Bill Dezellem with Tieton Capital Management. Please proceed. ...................................................................................................................................................................................................................................................... William Joseph Dezellem Chief Investment Officer & President, Tieton Capital Management LLC Q Yeah, thank you for taking the follow up. Two and I'll try to... ...................................................................................................................................................................................................................................................... A Hello. ...................................................................................................................................................................................................................................................... William Joseph Dezellem Chief Investment Officer & President, Tieton Capital Management LLC Q ...quickly. Number one is what were your renewal rate increases in the first quarter so far I guess the first two months of the year since you probably have that data? And secondarily, the Michigan reserve that was set one year ago, can you reiterate claims payments how those are unfolding relative to that reserve that was set one year ago? Please. ...................................................................................................................................................................................................................................................... A So for the reserve rate question the average rate increase that we took going into the end of 2017 was around 7% to 10%. Renewals are also run through our predictive model. And so we don't have a – we don't look at specific average premium per car versus last year, because depending on the predicted risk related to those cars, you could have more margin at a lower rate for a certain account versus a higher margin for – I'm sorry a higher margin and lower rate for certain account versus lower margin at a higher rate for other accounts. And so what we show on slide 4 is really the data we look at to show that how much absolute rate are we getting for an average account, which again was at the end of the year 2014 about 7% to 10%, so that would rolled into the first quarter. And then, what is the bias of the accounts we wrote towards the right side of that curve, that's the chart in the bottom right. And as you get further along to the right side, we are effectively capturing incremental margin off of our expected target. So there margin off of our expected targets. So there isn't a simple answer to that question Bill. But, you know, again we were basically, you know, putting 7 to 10 rate in, we should be getting more than that in terms of margin benefit, net of expected loss cost trend and we assumed expected loss cost trend last year to be about 3 percentage points to 4 percentage points. So, you know call the net rate increase. You know, three to five with maybe some incremental benefit from, from the effective better risk selection with the model. ...................................................................................................................................................................................................................................................... Q Thank you. ......................................................................................................................................................................................................................................................

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 35 Copyright © 2001-2018 FactSet CallStreet, LLC A So again. ...................................................................................................................................................................................................................................................... Q And Michigan. ...................................................................................................................................................................................................................................................... A So for Michigan, you know, again, you know, the, the work that we got did not provide a specific bucket of IBNR attributed directly to Michigan. It was contemplated, it was obviously a very heavily discussed you know with our actuaries, but you know, from a credibility perspective, you know, the overall result was factored into the total reserves. And so you know there wasn't a number last year that was reserve set aside for Michigan as it were. You know, it was incorporated into the overall recommendation. But what we, what I can tell you is, you know when we looked at the actual average pays for Michigan, you know, we saw relative stability in average pays on a year-over-year basis, whereas 2016 relative to 2015 we saw a big jump in average paid severity. And so that was really the thing that we were focusing on, because that was the expectation is that, and we said this in prior call, we didn't think average paid severity in Michigan was going to get better in 2017, but we didn't expect it to get materially worse and on a per claim basis it really didn't. ...................................................................................................................................................................................................................................................... Q And so when the reserve was set was the methodology set that you paid severity would be flat in 2017 or is that the backward looking view trend was increasing that it would increase? ...................................................................................................................................................................................................................................................... A Well, I think the view generally was that it would increase slightly. And again that is really what it did. Michigan clearly used up a disproportionate reserves over the last seven years versus what the earned premium from Michigan suggested it should. But to the tune of as I mentioned at the top of the call by about $38 million. But this year it was not Michigan getting much worse that created the issue. In fact if you look at the loss ratio for 2017, Michigan actually – with and without Michigan, in 2017 it does not appear that Michigan is hurting. We're still continuing to move towards having zero policies there. But – and that's a function of just taking so much rate any business we're writing is at an enormously higher rate than it was two years ago. But the bottom line is that the year end result in 2017 was a broad-based issue. It wasn't being driven by Michigan suddenly got twice as bad as everyone thought it would be. ...................................................................................................................................................................................................................................................... Q Great. Thank you for taking my follow ups. ......................................................................................................................................................................................................................................................

Atlas Financial Holdings, Inc. (AFH) Q4 2017 Guidance Call Raw Transcript 01-Mar-2018 1-877-FACTSET www.callstreet.com 36 Copyright © 2001-2018 FactSet CallStreet, LLC A All right. Sure. I appreciate the questions. ...................................................................................................................................................................................................................................................... Operator: Thank you. We have no further questions in queue at this time. I'd like to return the floor back over to management for closing remarks. ...................................................................................................................................................................................................................................................... Unverified Participant Thanks very much, Roya, and thanks for all of you. A lot of questions and hopefully it was an opportunity for us to begin to provide the transparency that we intend to continue to. Again, if there are any additional questions feel free to reach out to Paul or myself. I certainly look forward to seeing those of you who are going to be attending the APA Conference next week. We'll continue to provide additional information on the full earnings on our call, which as I mentioned is scheduled for mid-March and we'll also be participating at the CFA, New York Insurance Conference on March 19 and March 20th as well. So again thank you for your time and we appreciate you support. ...................................................................................................................................................................................................................................................... Operator: Thank you. This concludes today's teleconference. You may disconnect your lines at this time and thank you for your participation. Disclaimer The information herein is based on sources we believe to be reliable but is not guaranteed by us and does not purport to be a complete or error-free statement or summary of the available data. As such, we do not warrant, endorse or guarantee the completeness, accuracy, integrity, or timeliness of the information. You must evaluate, and bear all risks associated with, the use of any information provided hereunder, including any reliance on the accuracy, completeness, safety or usefulness of such information. This information is not intended to be used as the primary basis of investment decisions. It should not be construed as advice designed to meet the particular investment needs of any investor. This report is published solely for information purposes, and is not to be construed as financial or other advice or as an offer to sell or the solicitation of an offer to buy any securi ty in any state where such an offer or solicitation would be illegal. Any information expressed herein on this date is subject to change without notice. Any opinions or assertions contained in this information do not represent the opinions or beliefs of FactSet CallStreet, LLC. FactSet CallStreet, LLC, or one or more of its employees, including the writer of this report, may have a position in any of the securities discussed herein. THE INFORMATION PROVIDED TO YOU HEREUNDER IS PROVIDED "AS IS," AND TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW, FactSet CallStreet, LLC AND ITS LICENSORS, BUSINESS ASSOCIATES AND SUPPLIERS DISCLAIM ALL WARRANTIES WITH RESPECT TO THE SAME, EXPRESS, IMPLIED AND STATUTORY, INCLUDING WITHOUT LIMITATION ANY IMPLIED WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, ACCURACY, COMPLETENESS, AND NON-INFRINGEMENT. TO THE MAXIMUM EXTENT PERMITTED BY APPLICABLE LAW, NEITHER FACTSET CALLSTREET, LLC NOR ITS OFFICERS, MEMBERS, DIRECTORS, PARTNERS, AFFILIATES, BUSINESS ASSOCIATES, LICENSORS OR SUPPLIERS WILL BE LIABLE FOR ANY INDIRECT, INCIDENTAL, SPECIAL, CONSEQUENTIAL OR PUNITIVE DAMAGES, INCLUDING WITHOUT LIMITATION DAMAGES FOR LOST PROFITS OR REVENUES, GOODWILL, WORK STOPPAGE, SECURITY BREACHES, VIRUSES, COMPUTER FAILURE OR MALFUNCTION, USE, DATA OR OTHER INTANGIBLE LOSSES OR COMMERCIAL DAMAGES, EVEN IF ANY OF SUCH PARTIES IS ADVISED OF THE POSSIBILITY OF SUCH LOSSES, ARISING UNDER OR IN CONNECTION WITH THE INFORMATION PROVIDED HEREIN OR ANY OTHER SUBJECT MATTER HEREOF. The contents and appearance of this report are Copyrighted FactSet CallStreet, LLC 2018 CallStreet and FactSet CallStreet, LLC are trademarks and service marks of FactSet CallStreet, LLC. All other trademarks mentioned are trademarks of their respective companies. All rights reserved.