From: John Martello [john.martello@tower-research.com] Sent: Friday, March 26, 2004 4:12 PM To: rule-comments@sec.gov Subject: Summary of Intended Testimony -- Regulation NMS, File No. S7-10-04 To: rule-comments@sec.gov Summary of Intended Testimony -- Regulation NMS, File No. S7-10-04 John Martello Managing Director Tower Research Capital LLC I will testify that two of the key Regulation NMS proposals - the trade-through rule extension and the subpenny quoting ban -are flawed and will not accomplish the Commission's goals but in fact will have the adverse effects of making markets slower, less efficient, and less liquid, to the detriment of investors. In contrast, investors are best served if the trade-through rule is eliminated in all markets, and subpenny quoting is not only permitted, but encouraged. My specific points are outlined below. A. The trade-through rule creates an unfair advantage for slower market centers 1. Under the current trade-through rule as applied to the NYSE, the specialist has a free look-ahead period equal to the delay in the market. For example, if it takes 5 seconds to cancel an order on the NYSE, and the quote can be 5 seconds slow, if the market moves up after a marketable limit buy order is sent by a market participant at what was the market ask, then that order usually will not get filled. Conversely, if the market moves down after the same marketable limit buy order has been transmitted by a market participant, if the market participant attempts to cancel its buy, the specialist (who by virtue of his market incumbency has the fastest information feed) will usually fill the order before the cancel takes effect. 2. This look-ahead period can be most accurately represented by examining the optionality of a non-instantaneous market order as described in the Commission's Proposing Release 34-49325, and this optionality problem is the strongest argument as to why the Commission should not allow slow exchanges the same trade through rights as fast exchanges. As the Commission observed, a long delay in a market center implies that the person trying to cross the market is short an option for the time the order is outstanding. In effect, a customer placing a buy order on a slow exchange is in effect granting a free put option to the slow exchange. The optionality reflects a hidden charge for the person trying to execute at that order, and therefore in all cases that value of the put should be added to the slow market's price to make that price accurate, not just in market crossing scenarios (a cost of trading on slow markets which I believe most individual investors are completely unaware of). Therefore, if you were trying to figure out how optimally to place a buy order and an ECN had a price of $50.00 and a slow market maker had a price of $49.99, if the optionality cost is over $.01 then it is actually better to place the order on the automated market center at the worse price. Insisting on filling at the best available price without accounting for market delays therefore results in inherently worse fills for customer orders. 3. However, the Commission's attempt to estimate the cost of the option was incorrect. Allowing a fixed variance based on NBBO price as a way of accounting for optionality is inaccurate, and puts the government in the business of setting options prices in the market, which it should not be doing. The option has to be priced based on volatility, just like real-world options. For example, securities derived from broad market indices, such as QQQ or SPY, have inherently lower volatilities than the stocks that comprise those indices, and NASDAQ stocks generally have higher volatility than NYSE stocks. These differing volatilities result in vastly different costs of the option. Moreover, in periods of high volatility these option costs are higher still. However, under the proposed trade-through rule's minimum variation allowance, the Commission is essentially proposing a fixed price for options of a given price on these stocks. This illustrates a fundamental difficulty of having a trade- through rule - it has a lot of unintended consequences, which if they are fixable can only be fixed through even more detailed and complex regulation and enforcement. It would be reasonable to assume with the current high level of sophistication of the market that if these mandated option values are not perfectly accurate then the unsophisticated customer will be exploited by market participants capitalizing on an SEC-mandated arbitrage opportunity. 4. Non-automated execution facilities should be considered differently than automated, as the Commission noted in its request for comments on response times and degree of automation. Otherwise, an incentive will exist to create the slowest-responding ECN which give sophisticated traders using that ECN the greatest "look-ahead" period possible, and therefore the greatest value for their free option. Therefore, the Commission should be careful in how it determines what is an automated vs. non-automated execution facility. Response times should be included in the definition, and an average market order execution time of less than .75 seconds on average should be required for both cancel confirms and market order fills (the NYSE by our data takes 2.6 seconds for their auto-execution facility to execute). A .75 second market order execution time is very achievable with current technology as evidenced by the fact that it is a significantly slower fill time than what is average for all of the major ECNs. Further, the automated facility should be fully automated in all aspects, i.e., customers should be able to electronically place limit orders, the size of order should not determine whether the order is electronic or not, and the electronic market should operate even (and especially) in times of high volatility. None of these capabilities currently exist for the NYSE Direct+ system. Response times are even more important in times of very fast moving markets, and at these times many automated and non-automated facilities both slow down, which suggests that the Commission should set and enforce standards for robustness of response times in both normal and fast-moving markets. 5. I will also address the alternative proposal of the SEC, which is to allow professional traders to lock or cross markets when they have a special flag set in their order. This proposal seems to be in every conceivable scenario better for the non-professional investor than disallowing locked or crossed markets, though it is probably ideal to allow even non-professional investors to lock or cross markets due to the previously stated reasoning on optionality. If a professional investor crosses a slow market maker's offer with his bid, a non-professional investor seeking to sell the stock would be guaranteed a better price for the sell by the crossed amount, under all circumstances. A non-professional investor wishing to buy the stock would not be affected when compared to disallowing market crossing, under all circumstances. The only effect that mandating a maximum crossing amount will have is to lock in more profits for the professional trading community in general: by mandating a maximum price that a professional is allowed to place a bid, it in effect results in a lower bid relative to where the professional community believes the price to be and therefore a greater profit if an unsuspecting non-professional investor blindly places a sell order. Therefore, I believe there is a very strong argument that no limits on the degree of crossing should be in place, as it is only worse for the non-sophisticated investor. B. Subpenny quoting has measurably improved market efficiency 1. Much of the Commission's justification for the subpenny quoting ban is derived from out-of-date pricing and volume data, and should be reconsidered by the Commission using current market data. Problems attributed to subpenny quoting, such as lack of depth and "subpennying" of orders, have greatly diminished since the early period of subpenny pricing and are far less significant in the current subpenny market, as our studies of price and volume data show. The fast evolution of these markets also implies that even tighter spreads are achievable as the market steadily improves its efficiency. Limiting the tightest allowable spread to $.01, which is the average spread for many high-volume stocks today, legislates a maximum efficiency for the market instead of allowing further improvement. 2. Complaints by large block traders that subpenny quoting makes block trading more difficult should not outweigh the significant overall cost savings created by subpennies. Large block traders have sophisticated, low-cost auto- execution technology available to them that allows trading of large blocks with minimal price impact (guaranteed VWAP, best-efforts VWAP, smart order routing, etc.). 4. Subpenny trading creates demonstrably tighter spreads and lowers costs for investors. Investors who pay the spread (by placing marketable limit orders or market orders) have a cheaper fill rate under a subpenny regime. This is especially relevant for low priced stocks (under $10), and very liquid stocks where spreads are pegged at a penny. I will discuss analyses I have performed of several widely-held securities which demonstrate that the aggregate price savings to investors of subpennies is in fact economically significant - in the hundreds of thousands of dollars per day for certain highly liquid stocks.