The Cost of High-Frequency Traders and Latency Arbitrage

A new study released by the Financial Conduct Authority (FCA) in the U.K. has reignited concerns around the ability of predatory high-frequency traders (HFTs) to profit on equity trades by using a technique called “latency arbitrage.” Latency arbitrage is a practice used by HFTs that utilizes a fragmented market structure and speed to “snipe” stale quotes on one equity exchange while simultaneously transacting on another exchange to lock in a “risk-free” profit. Effectively, HFT firms are able to exploit the delays in changing quotes for securities while information is being disseminated into the marketplace, profiting from the arbitrage opportunity that is created by this information latency. This article from the Wall Street Journal highlights some of the key findings from the report, citing that while the “tax” of latency arbitrage is low on a per transaction basis, applying that to global trading volume could potentially amount to just under $5 billion in economic rent. Furthermore, that economic rent in the form of a latency tax is concentrated in the hands of a small number of HFTs. These are the firms who are able to build the fastest infrastructure (either through code, location of servers, transmission mechanisms, etc.) and the resulting profits make it possible for these firms to continue the “arms race” of investing in additional high-speed infrastructure.

The most interesting part about this study is that previous attempts to quantify the effect of predatory practices by HFTs have been inconclusive due to the opaque nature of the strategy and the fact that only limit order data was gathered. This new study by the FCA uses messaging data where attempts to trade can now be measured, providing researchers with more reliable data on isolating the latency arbitrage “races” to better measure volume and the financial impact of the strategy.

The rebuttal to this piece from the HFT community would be that the study only looked at a small sample size of data on one marketplace, making it a stretch to extrapolate those “tax” figures to global equity trading volume. That argument does have some validity, given that only 22 percent of the total trading volume in the study was a function of latency races. However, in this case, the value of the economic rent from the races doesn’t change, it’s just whether this strategy could be reliably applied to total trading volume across global equity markets. The answer to this question is probably no, causing the total estimated economic rent tally to likely be overstated, but it does highlight concerns for the trading community as it relates to overall marketplace liquidity. While the “cost” of latency arbitrage is more impactful for large institutional market participants than retail investors, the concern is that the increase in transaction costs could reduce the amount of passive “resting” orders that provide natural liquidity to the market, creating a positive feedback loop that is detrimental to overall liquidity conditions and ultimately pushes up execution prices. Although advances in technology and information systems have on balance been a positive boon for all investors as it relates to liquidity and execution, it is always important to consider the evolving nature of the investment landscape and how it might impact you in unique ways.