Subject: File No. S7-07-04
From: Phillip S Martin

April 7, 2004

I understand that you are currently reviewing the practice of internalization of orders in the options industry. I worked as an options trader on the floor of the American Stock Exchange from 1986 until 1999. During that time, I was frequently amazed at the creative ways in which brokerage firms managed to exploit the legality of internalization to enrich themselves at their own customers expense. I should like to call your attention to one practice in particular that demonstrates the danger of allowing brokerage houses to act simultaneously as both principal and agent.

First, I need to provide some background on the mechanics of trading an option spread. For the sake of an example, lets say that two options on a given underlying, which we shall call option A and option B, are currently being quoted 1-1 3/8 and 2 1/8-2 respectively. Suppose that a customer wishes to trade a spread in these options. He might wish either to buy the spread i.e., buy option B and sell option A or to sell the spread i.e. sell option B and buy option A.

If he were to trade each option individually, he would receive if he were to sell the spread selling B at 2 1/8 and purchasing A for 1 3/8 and he would pay 1 if he were to buy the spread. Apparently, then, market in the spread is - 1 .

I say apparently because, in fact, the market is tighter than that. Traders are generally willing to trade between the bid and offer on one leg of a spread. In fact, since a spread is inherently less risky than a single option, they are often willing to make a tighter market on a spread than they would on either individual option. In this case, they might well make the market on the spread 1 to 1 1/4. Clearly, before trading a spread, a customer should always ask for the spread market rather than simply infer the spread market from the individual options.

But many customers, even those who trade in large volume, are apparently unaware of this fact. Further, it is important to note that spread markets are not disseminated. Even though, in this case, traders might be willing to pay 1 for this spread or to sell it at 1 , there is no way for a customer to know this other than to ask his broker to obtain a market in the spread. In practice, customers often give their brokers spread orders without asking for a market first.

Consider first what should happen if a customer wishes to sell 500 of these spreads. Ideally, he would first ask his broker to obtain a market. The broker would go into the crowd, ask for a market, and obtain the market of 1 to 1 , with a size of, say, 200 by 200. The broker would then report this market to his customer. Lets say the customer then gives the broker an order to sell 500 spreads at 1. The broker would then return to the crowd with the order. The crowd would pay 1 for 200 spreads, then would show the broker a new bid, say 7/8, for the remaining 300 spreads.

At this point, a variety of things might happen. The customer might choose to leave his order active, or to sell the remaining 300 spreads at 7/8, or to cancel the balance of his order. Another possibility is that the brokerage firm might choose to facilitate the order. In an effort to keep their customer happy, they might choose to pay 1 themselves for the balance of the order.

To facilitate the order in this way is to act simultaneously as principal and agent, which, as we know, creates a potential for a conflict of interest. In this case, the action is legal under the facilitation rule, which allows firms to take the other side of their customers orders provided no one else is willing to do so. Note that this is the original form of the facilitation rule. Since then, the rule has been modified many times, providing additional exceptions and making it easier, not harder, for the firm to act as both principal and agent. To keep the discussion simple, Ill continue to assume the original form of this rule. This original, apparently harmless, rule was in fact noxious enough in practice, and the various modifications over the years have only made it more so.

To see how application of this rule can harm the customer, consider another example. Suppose a less savvy customer doesnt bother to quote the spread first. He simply gives his broker an order to sell 500 spreads at . By the rules of the exchange, what should happen is this: As in the first example, the broker should come into the crowd and ask for a market in the spread. The crowd would quote a market of 1 1 , 200 by 200. Again, the broker should offer 500 spreads at 1, and, again, he would immediately sell the 200 the crowd had bid for. The crowd would then show him a 7/8 bid for the remaining 300, and the broker would probably sell the balance at that price, knowing the customer will be happy to have done better than he intended to.

The firm, of course, knew that the customer was willing to sell the spread as low as . They also knew that 3/4 was a very cheap price. Accordingly, they would like to buy some of the spreads themselves at this price if possible. Legally, this is not possible. First of all, the bid for 500 spreads is higher than . Second, there is adequate liquidity so there is no reason for the firm to take the other side of the trade. Third, in their role as agent, they should want to sell the spread for as much as possible. To allow the spread to spread at would be a clear violation of their duty to their customer. Despite these hurdles, it did not take the firms long to figure how to trade the spread at anyway, buying some of it themselves, by using the facilitation rule.

What would happen is this: The broker would come into the crowd and ask for a market in the spread. The crowd would quote a market of 1 1 1/4, 200 up. The broker would then say, I have 500 spreads at , but Ill sell them only if I can buy half of them for the firm. In other words, the broker would offer a quid pro quo. He would allow the crowd to buy the spread at even though they had shown him a 1 bid. In return, he would ask the crowd, in effect, to pretend that they were willing to buy only 250 at 3/4. The firm could then facilitate the balance of the order, buying the remaining 250 for themselves. If the crowd did not cooperate, the spread would simply not trade. Although I cant say for sure what happened, I assume that the firm would then simply tell their customer that they couldnt find anyone to pay for 500 spreads. Thus they would sacrifice that order to make sure crowd cooperated the next time.

As options started to be listed on multiple exchanges, this practice became more and more frequent. Now, if a crowd would not cooperate, the firm would not have to sacrifice the order. They would simply take the order to a different exchange. The fact that spread markets were not disseminated made this possible. The firm could sell a spread at on one exchange, buying half of them for themselves, knowing full well that the spread was 1 bid on a different exchange.

Firms will, of course, deny this practice. But I can assure you that by the time I left the floor in 1999 it was a daily occurrence. The point I wish to make is that allowing a firm to act as both principal and agent creates opportunities for abuse. Trying to control this abuse by placing restrictions on when it can happen does not work. The firms will always find a way to circumvent the restrictions. You can put an end to their placing their own interests over those of their customers only by insisting that they can never act as both principal and agent.