The data-driven demise of the flash boys myth

September 23, 2016 01:00 PM
Recently, a study by two professors at the University of California, Berkeley, tested the idea that high-frequency trading (HFT) strategies use information from fast data feeds to front run investors with slower data feeds.

Recently, a study by two professors at the University of California, Berkeley, tested the idea that high-frequency trading (HFT) strategies use information from fast data feeds to front run investors with slower data feeds. This was the central claim in the book “Flash Boys: A Wall Street Revolt” and the impetus for the creation of the Investors Exchange (IEX) dark pool, which has recently been approved as a national stock exchange. 

Turns out, the professors found no evidence that supports the claim. Or as one of the study’s authors put it more succinctly in a Reuters article, “That’s not happening.”

Specifically, the study abstract notes, “Trading surrounding Securities Information Processors (SIP)-priced trades shows little evidence that fast traders initiate these liquidity-taking orders to pick off stale quotes. These findings contradict claims that fast traders systematically exploit traders who transact at the SIP (the national best bid and offer).” 

It’s been more than two years since a widely-watched segment on 60 Minutes declared the market is rigged by speedy traders front running others. It’s a claim that has been repeated as fact many times since then and served as the basis for inconclusive hearings and probes, as well as dismissed lawsuits. But it lives on.

Fear of “latency arbitrage” and “front running” are the reasons for the existence of IEX. In two comment letters, sent 10 days apart, IEX explained to regulators that it’s then pending exchange application needed an exemption to exchange rules “to counteract the more pernicious aspects of speed-based trading.” Neither letter offered supporting data to substantiate the behavior they sought to counteract, or quantify its effect. It just repeated the ominous phrase, in bold typeface, both times. This is in stark contrast to letters opposing IEX, which included specific examples and data-driven analysis. 

In light of the data-driven study by the U.C. Berkeley professors, has all of this rancor been unnecessary? Has the added complexity of granting an exchange an intentional delay, and now at least two others considering it, been unwarranted? Has the demonization of high-frequency trading that accompanied accusations of abusive, speed-based front running, been unfair?

Unsurprisingly, the usual rigged/broken/conspiracy crowd has desperately tried and failed to debunk this study. First, by claiming the data was wrong; it isn’t. And then by focusing on nearly non-existent occurrences of price difference. And while the study does find that a tiny (0.062%) of trades left a liquidity taker in a worse position, it also says that is “primarily a product of chance rather than of HFT design.” 
Hardly something that justifies turning the principles of our National Market System upside down to combat.

One prominent critic even dismissed the study as “a SIP/direct feed study without actually taking direct feeds.” But he misses the point. This is a SIP versus participant timestamp study, and there is no faster data than a participant timestamp that gives the price at the precise time of the inception of the trade with no data communication latency whatsoever. Lightning fast direct feeds are pokey in comparison. So the study poses the hypothetical question: If you used the slowest data feed available (the SIP) and I used the fastest data possible (participant timestamps), would I be able to front run you with impunity? The answer is, “No.” 

At least not on purpose.

It is probably impossible to quantify the damage that the “rigged markets” campaign has inflicted on investor confidence. But what can be quantified is that the S&P 500 Index has risen about 15% (18% when including reinvested dividends) since the airing of that 60 Minutes piece. Anyone scared out of the market didn’t collect that money. And that’s a tragedy for retirement-minded investors who can ill-afford to miss large gains in the stock market. 

Investors are owed an apology. It is one thing for a firm to claim they offer investors better execution, better customer service, or simpler pricing. But quite another to foment fear by saying that all their competitors are colluding to rig the market.

Clearly, there is no substitute for good data. Not popular books, provocative exchange applications or passionate letters. This study should serve as an example of why we need data-driven, holistic analysis of market structure problems before we start introducing “remedies” that further complicate the trading ecosystem.

About the Author

Bill Harts is CEO of Modern Markets Initiative, an advocacy effort organized by some of the industry’s leading quantitative trading firms. @modernmarkets