Do Inverted Venues Leak Information?

At some point in your career, I am sure you have heard the claim that posting orders on inverted exchanges leaks information. Because the limit order trader implicitly conveys their willingness to pay a fee – and forego a rebate – when posting to an inverted venue, the argument goes, she is signaling to the market that she is more impatient and/or less price-sensitive than other traders. Such a signal could lead to price impact as traders on the opposite side of the market respond by pricing their orders less aggressively and traders on the same side respond by pricing their orders more aggressively.

Conceptually, this argument makes sense. Routing to an inverted venue is similar in spirit to increasing the bid price (or decreasing the offer price) in that the limit order trader is making her net price (i.e., inclusive of fees) more aggressive. But the question is really “how much incremental information is there when a trader posts to an inverted venue”? While the net price is relatively more aggressive, the trader has not gone so far as to make her gross price more aggressive, by either bettering the best price in the market or by crossing the spread. Furthermore, the trader may simply be routing to an inverted venue as a rational response to long queues in the market. If, for example, a trader routes to an inverted venue to “jump the queue”, perhaps joining others already posted on inverted venues, how much incremental information is in that order?

But even if the incremental information causes price impact, this alone is not enough to justify avoidance of inverted venues. Rather, any incremental price impact needs to be weighed against the potential benefits of inverted venue usage. For example, inverted venues may have higher fill rates and a shorter time-to-fill since marketable orders have an economic incentive to route to inverted (i.e., rebate-paying) venues over non-inverted venues. Are these benefits valuable enough to overcome the potentially higher information leakage? At its heart, this is an empirical question. Fortunately, two well-researched studies have shed light on this issue.[1]

The Empirical Evidence

The first study done by Pragma Trading shows the relative performance of orders routed to inverted venues relative to non-inverted venues based on “horse races”, where they randomly allocated VWAP child orders to different venues and monitored the initial price impact (i.e., prices rising in response to the initial submission), the effective spread capture, and the post-fill performance. They then compare the average performance across inverted and non-inverted venues.[2]

The authors find that there is, in fact, an initial price impact that occurs within the first 5 seconds after the order is displayed on an inverted venue. However, the magnitude of the price impact is effectively the same as that which occurs on a non-inverted venue. At the same time, the average fill performance is better on inverted venues. This gives empirical evidence that, not only do inverted venues not have more price impact than non-inverted venues, they have superior execution performance overall.

Such a finding is even more meaningful when one considers that the Pragma study involved naively routing to an inverted venue or to a non-inverted venue. This is an important point since orders sent to inverted venues are often routed there tactically, e.g., when the queues are already deep (and often deep on both sides), making the incremental information of an inverted order relatively small. But the Pragma study goes a step further. They route to inverted venues randomly, without regard to the state of the book. So, if anything, their results overstate the price impact and understate the fill performance of tactical usage of inverted venues. I suspect their results would be even stronger had they modified their study to first “optimize” their inverted usage and then randomly deviate from that strategy by routing some orders ideally suited for inverted venues to the “next best” non-inverted venue.

A second study – done by Professors Robert Battalio, Shane Corwin, and Robert Jennings –investigates whether there is a correlation between limit order performance and exchange fees.[3] Their analysis is done using a proprietary dataset from a major broker dealer’s algorithmic child orders as well as using publicly available TAQ (i.e., historical SIP) data. Consistent with the Pragma findings, the authors find that there is, in general, an inverse relationship between fill quality and fees, with non-inverted venues performing worse than inverted venues.[4] Specifically, they find that the “realized spread”, which is defined as the spread capture net of any post-trade price moves (i.e., adverse selection) is better on inverted venues than on non-inverted venues.

Conclusion

In summary, then, the answer to the title question appears to be “no”. Not only do inverted venues not incur significantly more price impact, they also provide superior execution performance than non-inverted venues. Of course, as noted in the Battalio-Corwin-Jennings study, the relative benefits depend on factors like volatility, price level, etc. But these studies put the onus on brokers to prove that biasing their passive trading toward high-fee venues is actually adding value and not simply a way for brokers to monetize their client’s flow.

Footnotes

[1] A third paper, written by Hernan Otero, Alak Vasa, and me, also documents a relationship between inverted venue use and improved limit order performance. The paper was published in the now-defunct Journal of Trading, but is also available at here. The study’s data is over 10 years old, but the results we uncovered were consistent with the two studies outlined here.

[2] Inverted Exchanges: Higher Leakage or Higher Quality?, Pragma Trading, March 2018, available on request here.

[3] Their paper has been published in the Journal of Finance, one of the top academic finance journals, and is available (behind a pay wall) here. A working paper version can be found here,

[4] There was one notable exception. The authors find that the NYSE actually fared better than its fee ranking would otherwise suggest.

The author is the Founder and President of The Bacidore Group, LLC. For more information on how the Bacidore Group can help improve trading performance as well as measure that performance, please feel free to contact us at info@bacidore.com or via our webpage www.bacidore.com.