NASDAQ: AFH Preliminary 2017 Fourth Quarter and Year End Conference Call March 1, 2018

Forward Looking Statement 2 Statements in this presentation, including the information set forth as to the future financial or operating performance of Atlas Financial Holdings, Inc., American Country Insurance Company, American Service Insurance Company, Inc., Gateway Insurance Company and/or Global Insurance Company of New York (collectively, “Atlas”), that are not current or historical factual statements may constitute “forward looking” information within the meaning of securities laws. Such forward looking statements involve known and unknown risks, uncertainties and other factors which may cause the actual results, performance or achievements of Atlas, or industry results, to be materially different from any future results, performance or achievements expressed or implied by such forward looking statements. When used in this presentation, such statements may include, among other terms, such words as “may,” “will,” “expect,” “believe,” “plan,” “anticipate,” “intend,” “estimate” and other similar terminology. These statements reflect current expectations, estimates and projections regarding future events and operating performance and speak only as to the date of this presentation. Readers should not place undue importance on forward looking statements and should not rely upon this information as of any other date. These forward looking statements involve a number of risks and uncertainties. Some of the factors facing Atlas that could cause actual results to differ materially from those expressed in or underlying such forward looking statements include: (i) market fluctuations, changes in interest rates or the need to generate liquidity; (ii) access to capital; (iii) recognition of future tax benefits on realized and unrealized investment losses; (iv) managing expansion effectively; (v) conditions affecting the industries in which we operate; (vi) competition from industry participants; (vii) attracting and retaining independent agents and brokers; (viii) comprehensive industry regulation; (ix) our holding company structure; (x) our ratings with A.M. Best; (xi) new claim and coverage issues; (xii) claims payments and related expenses; (xiii) reinsurance arrangements; (xiv) credit risk; (xv) our ability to retain key personnel; (xvi) our ability to replace or remove management or Directors; (xvii) future sales of common shares; (xviii) public company challenges; and (xix) failure to effectively execute our business plan. The foregoing list of factors is not exhaustive. See also “Risk Factors” listed in the Company’s most recent registration statement filed with the SEC. Many of these issues can affect Atlas’ actual results and could cause the actual results to differ materially from those expressed or implied in any forward looking statements made by, or on behalf of, Atlas. Readers are cautioned that forward looking statements are not guarantees of future performance, and should not place undue reliance on them. In formulating the forward looking statements contained in this presentation, it has been assumed that business and economic conditions affecting Atlas will continue substantially in the ordinary course. These assumptions, although considered reasonable at the time of preparation, may prove to be incorrect. When discussing our business operations, we may use certain terms of art which are not defined under U.S. GAAP. In the event of any unintentional difference between presentation materials and our GAAP results, investors should rely on the financial information in our public filings.

Impact of Reserve Strengthening Forecast for 2018 3 Reserve Changes • Strengthening for older accident years • More current years stable Expected BVPS at December 31, 2017 • $7.25 - $8.00 • Includes approximate $0.55 reduction in Deferred Tax Asset due to tax policy change, unrelated to strengthening Forecast for 2018 • $300 million+ in GWP, subject to actions by rating agencies • At that level of premium and with a Combined Ratio in mid to high ‘80s, it is reasonable to expect annual net EPS to exceed $2.00

Pricing Activity And Impact of Predictive Modelling 4 -10 -5 0 5 10 15 20 25 Q 1 1 2 Q 2 1 2 Q 3 1 2 Q 4 1 2 Q 1 1 3 Q 2 1 3 Q 3 1 3 Q 4 1 3 Q 1 1 4 Q 2 1 4 Q 3 1 4 Q 4 1 4 Q 1 1 5 Q 2 1 5 Q 3 1 5 Q 4 1 5 Q 1 1 6 Q 2 1 6 Q 3 1 6 Q 4 1 6 Q 1 1 7 Q 2 1 7 Q 3 1 7 Q 4 201 7 Atlas vs ISO Average ISO Recommendation (for period) Atlas Rate Change (Ave)

Strengthening by Business Segment 5 -$5.0 $0.0 $5.0 $10.0 $15.0 $20.0 2011 & Prior 2012 2013 2014 2015 2016 2017 Mill io n s ASI Pool Reserve Change by Accident Year (net, after tax) MI Only PA No-MI LI No-MI TA No-MI All Other no-MI -$5.0 $0.0 $5.0 $10.0 $15.0 $20.0 2011 & Prior 2012 2013 2014 2015 2016 2017 Mill io n s Global Liberty Insurance Company of New York Reserve Change by Accident Year (net, after tax)

Fact Pattern Positive Changes in Business Which Challenge Near Term Actuarial Work 6 2015 • Atlas initiated machine learning based predictive analytics as underwriting tool 2016 • Predictive analytics directly integrated into Point of Sale system (AtlasX) and rolled out to agents during year • On June 1, predictive analytics integrated into claims process with the objective of identifying potentially large claims earlier the previously possible with the goal of amplifying Atlas’ core strengths as follows: • Meritorious claims could be settled more quickly with less indemnity “leakage” and allocated loss adjustment expense (“ALAE”) • Defense and/or mitigation of non-meritorious claims could be engaged earlier in the process to effective better outcomes 2017 • As a direct result of the above initiatives, significant compression was observed in closure rates on large losses (page 9) • Paid amounts increased in absolute terms but are decreasing as percentage of net earned premium (page 10) • Claim inventory relative to in-force cars insured declined (page 11)

Heavier Reliance on IBNR Pre-Model 7 Transition to Predictive Model Based Case Reserving • Under predictive model based case reserving, each claim should be set to ultimate at 45 days of claim age • Reliance on IBNR for to estimate ultimate at year end until AY 2016 • Case reserves for claims too old to model were set at a factor based on average incurred at that time • Comparisons of paid modeled claims versus case demonstrates redundancy ITD 0 2,000 4,000 6,000 8,000 10,000 12,000 1 3 5 7 9 11 13 15 17 19 Average Case Reserve on Open Liability Claims Before After Months following claim receipt

Savings Observed Post Modelling 8 Indicated Savings of 10% - 17% on similarly scored claims (before and after implementation of predictive model based claim triage) Change observed in allocated loss expense (e.g. defense costs, etc.) also supports view that most challenging claims are being extinguished earlier

Compression Resulting From Predictive Model Based Triage/Routing 9 Large percentage of small claims, relatively unchanged Notable shift in settlement time for mid size claims Most significant shift related to larger severity cohort of claims These are the groups intentionally targeted using predictive analytics to accelerate closure or defense earlier in process Notes • “Before” includes claims reported during the 90 days immediately before implementation of predictive analytics in claim triage/routing and “After” is a 90 day period starting three months after implementation • Each line shows monthly development for 3 months of reported claims in each group • Denominator is all claims reported in period regardless of open or closed for any amount 0.0% 0.5% 1.0% 1.5% 2.0% 2.5% 3.0% 1 3 5 7 9 11 13 15 Claims $25,000 - $50,000 Before After 0% 1% 2% 3% 4% 5% 6% 7% 8% 9% 1 3 5 7 9 11 13 15 Claims $10,000-$25,000 Before After 0% 20% 40% 60% 80% 1 3 5 7 9 11 13 15 Percent of Total Liability Claims Closed Claims < $10,000 Before After

Average Paid Severity By Size Cohort 10 Smaller claims on average have been settling for less Average paid severity on larger claims is not inflating despite earlier closure rate Note: claims in these cohorts are 20X – 30X the large number of smaller claims Notes • Includes all claims closed for the severity ranges show at 15 months, including zero indemnity claims • Paid Loss + Paid ALAE / Closed Claims including partial payments • Each line shows monthly development for 3 months of reported claims 0 500 1,000 1,500 2,000 1 3 5 7 9 11 13 15 Average Paid Loss & ALAE Severity on Closed Liability Claims (With Indemnity, Cumulative) Closed Claims < $10,000 Before After 0 5,000 10,000 15,000 20,000 25,000 1 3 5 7 9 11 13 15 Closed Claims $10,000-$25,000 Before After 0 10,000 20,000 30,000 40,000 50,000 1 3 5 7 9 11 13 15 Closed Claims $25,000 - $50,000 Before After

Claim Closure Patterns 11 Rolling 12 Month Claims Presented Open Inventory Inflection Point Evidence of Improved Risk Selection Incoming claim counts are flat despite 20% yr/yr growth In addition, accelerated closure rates are clearing inventory more quickly Traditional actuarial methods are trend based, potentially muting impact of inflection point

Adjusted Loss Ratio 12 Potential Unrecognized Improvements From Predictive Analytics Risk Selection (~4%) Severity Reduction (~10%) Accident Year Loss & LAE Ratio 2012 92.8% 2013 73.7% 2014 66.4% 2015 70.9% 2016 67.5% 2017 67.0% Global Liberty Accident Ye r Loss & LAE Ratio 2012 91.7% 2013 95.5% 201 91.3% 2 15 77.5% 2016 63.9% 2017 57.9% ASI Pool Companies Accident Year L ss & LAE Ratio 20 2 86.5% 2013 85.9% 20 4 76.6% 2015 68.5% 2016 59.5% 2017 58.0% ASI Pool Companies (excl MI) Includes Loss & ALAE only Includes ULAE Includes ULAE Includes ULAE

Prior Actions Taken to Address Relative Underperformance Michigan 13 Prior sequential rate increases actuarially determined to achieve expected profitability targets • Maximum rate supported taken each year 2014: 30% 2015: 25% 2016: 60% 2017: 60% • Year end actuarial work supported incremental rate increases • Prior to 2016, actuarial pricing work indicated additional rate could achieve expected profitability • Increased severity observed in 2016 resulted in decision to strengthen reserves as well • Claim feature inventory limited to 420 by year end 2017 (of which 259 have been scored) • Exposure reduced to less than 1% by year end 2017 and continuing to decline 0 500 1000 1500 2000 2500 3000 Jan-1 6 Fe b -1 6 Mar-1 6 A p r-1 6 May -1 6 Ju n -1 6 Ju l-1 6 A u g-1 6 Se p -1 6 O ct -1 6 N o v-1 6 D ec -1 6 Jan-1 7 Fe b -1 7 Mar-1 7 A p r-1 7 May -1 7 Ju n -1 7 Ju l-1 7 A u g-1 7 Se p -1 7 O ct -1 7 N o v-1 7 D ec -1 7 Jan-1 8 Michigan Inforce