IEX Petition to SEC Could Create Two Tiered Exchange Model

Hello, it’s me… (Mr. Lewis)

Michael Lewis’s letter to the SEC on the IEX exchange application reminds me of the Monty Python sketch, “Dennis Moore.” The world’s most inept highwayman, Moore tries his hand at the Robin Hood schtick by stealing “lupins” from the rich and giving them to the poor. He is finally schooled by the sick and dying poor who’ve had enough of the flowers, “Why don’t you go out and steal something nice, like Venetian silver.”

Our hero rides off and starts stealing more useful things, candelabras and tiaras. He keeps pilfering from the rich until the poor are rich and the rich are poor. By the end of the story, the “hero” is robbing from the poor and giving the spoils back to the rich…

In the Lewis letter he portrays IEX as the defender of the average investor and ignores the substantive issues raised by other commenters, including the blatant attempt by IEX to create a competitive advantage for their own affiliate. He asserts that IEX stands for “a spirit of fairness” and is a solution to the “problem of predatory trading.” Nice words, if only they were true…

As I’ve said before, the IEX structure, if approved “as is,” would usher in a new Two-Tiered Market. The precedent of allowing an exchange to place its own routing broker in an advantageous position, would create a competitive disadvantage for all commercial routing outside of exchanges. This would severely limit the ability of broker dealers to have control over their best execution obligations and be a major setback for the markets. In a very real sense, it would be taking money from brokers and investors to provide a regulatory advantage to IEX. Also consider that IEX would be the most expensive exchange if it were approved today. It has a profit margin of $0.18 per hundred shares traded compared to less than $0.05 for the other exchanges. If they were allowed to leverage their routing advantage to increase market share, that would indeed be a wealth transfer to the rich…

I have often thought that the “benefit” of the speed bump is to investors that post orders on other exchanges and not to those that post on IEX. As such, while it is not as silly as giving “lupins” to the poor, it is also not what they claim it to be.

IEX was created as a solution to a perceived structural problem with the equity market. That problem, which used to be widespread, was an inability to buy or sell the total amount of stock displayed at a posted price. For the most part, this occurred because routing software was being run on poorly architected networks. This meant that the router, despite sending orders to multiple exchanges simultaneously, delivered those orders with large gaps in time between. Lewis’s book “Flash Boys” showed that if such gaps are too large, then traders could possibly “race” the slower orders and either trade ahead of them or cancel the quotes that the orders are trying to access1.

Today, most large broker dealers have built smart order routing technology on well architected networks. They are typically co- located at one of the three main datacenters used by the exchanges. (NASDAQs Carteret, ICE’s Mahwah, or the Equinix NY4/5 facility used by BATS). As a result, it is highly unlikely that the majority of routers leave a sufficient gap between the times their orders are received by the exchanges, to make a “race” possible. It is, of course, a good idea for all clients to ask their brokers for reporting on the efficiency of routing. Analytic reports that measure the fill percentage of orders, compared to the displayed size in the market, is the best method to show if there are any “race” conditions occurring.

Even if it were true, that a lot of order routers leave gaps in between the orders they are aggressively routing, it is hard to understand why that would encourage investors to post on IEX. While it is clear that the IEX speed bump will make gaps less likely to occur when IEX is accessed first, it only helps the posted orders on the other exchanges. It is literally irrelevant to the investor who posted an order on IEX if orders posted on other exchanges are easier to trade with. The only IEX clients that actually benefit from the speed bump are the ones who either aggressively trade with those resting orders or utilize the IEX router.

Apologies to Mr. Lewis, but even if the order delivery time gaps were still a significant problem, the IEX speed bump would only help investors if universally adopted. That, however, is unlikely, as some exchanges actively compete for posted orders and use their speed as an attraction. Some believe that faster markets deliver a higher certainty of being filled, while slower exchanges with less accurate quotes provide less certainty. This is critical for investors posting limit orders since fill probability is one of the most important reasons to choose a venue. Analysis of both fill probability and the price movement, after an order is filled, is key to venue selection. Venues that tend to see significant adverse movement whenever orders are filled can be quite costly to use. That, much more than rhetoric, should be how investors evaluate IEX. If they were to deliver high fill probability and low adverse price movement, they would be a welcome addition to the US equity market landscape.

While I agree that IEX should be allowed to become an exchange with their speed bump intact, it most certainly should not be approved with the ability for orders to go to their own router without going through the speed bump first. The SEC should not allow this decision to be driven by the populist considerations of Mr. Lewis, or anyone else. It is important for the regulators to keep our markets fair and avoid giving any single participant a competitive advantage, especially one with marketing savvy like IEX.

VOICEOVER: Well it may be the end of that, but it’s certainly far from the end of – well in fact it’s the beginning – well not quite the beginning – well certainly nearer the beginning than the end – well yes damn it, it is to all intents and purposes the beginning of this year’s Ideal Loon Exhibition

1This occurs if the latency difference between the slowest order and the fastest delivered order from a router exceeds the turnaround time of the fastest order by enough to allow a trader to race it. In the book, Mr. Lewis described a router based in NYC that took a much shorter time to get to Weehawken (where BATS used to be) than to reach either Carteret or Mahwah. In his example, the gap was sufficiently large for a fast trading system in Weehawken to see BATS market data and send an order to the other exchanges before the original order arrived.

David WeisbergerDavid Weisberger is the Managing Director and Head of Market Structure Analysis of RegOne Solutions, by Markit