Over the summer, while college students interned, families vacationed, and the world watched the Olympics, several teams of academics put trading activity under a microscope and came up with some interesting observations about today’s electronic marketplace.
Researchers from the University of California at Berkeley, Vanderbilt University, and the Illinois Institute of Technology published two separate papers that examine the role of high frequency trading (HFT) and its effect on certain aspects of market quality. The research papers ultimately rejected claims that the use of HFT technology has led to “rigged” markets with “phantom” liquidity.
If you missed them when they first came out, or are simply too busy to read them, here are the FIA PTG CliffsNotes to get you caught up.
Robert P. Bartlett III, and Justin McCrary from the University of California at Berkeley investigated whether fast traders are taking advantage of the time differences between the market data published by exchanges and the National Best Bid or Offer (NBBO) reported using consolidated market data by Securities Information Processors, or SIPs. The question they attempt to answer is whether fast traders get a head start because of their use of the data in the faster direct exchange feeds.
Bartlett and McCrary tested this contention by analyzing data showing the exact timestamp, measured in microseconds, on each update published by the SIPs. They used this dataset to calculate the quote and trade latencies for all individual stocks in the Dow Jones Industrial Average. Next, they analyzed whether traders using the slower SIP price information would have fared better if they had used direct feeds.
Their study, entitled, “How Rigged Are Stock Markets? Evidence from Microsecond Timestamps,” is the first systemic analysis of the latency with which the SIPs process quote and trade data.
The result? “Overall, our analysis contradicts the conventional wisdom that slow, liquidity-taking traders are systematically harmed when they have their trades priced at the slower SIP NBBO.” Specifically, the authors’ found that on a value-weighted basis liquidity-taking trades in the sample gained on average $0.0002 per share by having their trades priced at the SIP NBBO rather than the Direct NBBO.
While liquidity-taking trades benefit on average when priced at the SIP NBBO, these trading gains come at the expense of liquidity providers who purchase shares at prices that are higher than reflected in the Direct NBBO (or sell shares at prices that are lower than reflected). Using the new timestamp data to explore trades surrounding these “mispriced” trades, the authors find virtually no evidence that these trades are the result of fast traders using market orders to “pick off” slower market participants. The conclusion is clear: the gains and losses between market participants using slower feeds versus faster feeds “appears to be primarily a product of chance rather than of HFT design.”
The second study published this summer comes from Jesse Blocher of Vanderbilt University, Ricky Alyn Cooper and Ben Van Vliet of the Illinois Institute of Technology, and Jonathan Seddon of Audencia Nantes School of Management. Their study, entitled “Phantom Liquidity and High-Frequency Quoting,” was published in the Summer, 2016 edition of The Journal of Trading. They examine a question first raised by the SEC about whether HFT provides so-called “phantom” liquidity which “disappears before longer-term traders can access it.” To determine whether or not liquidity is real or merely phantom in nature, they analyze “cancel clusters” which “arise from the combined activity of HFTs cancelling their limit orders within a small timeframe, due presumably to common private information.”
They examine S&P 500 Index stocks during 2012 and identify clusters of extremely high and extremely low limit-order cancellation activity. They found that cancel clusters are not a dominant feature of the trading day. They begin and end rapidly, lasting for only 5.68 seconds and occupying only 6.7% of the trading day. Moreover, they found that executions are comparatively rare during cancel clusters, and most trading execution takes place outside of these clusters. When cancellations are low, prices are only determined by executions moving the price. “This means that investors are paying for the price discovery in executions at prices that are immediately changing,” the authors explain.
The authors conclude that HFT “liquidity providers have simply replaced low frequency market makers with lower cost price discovery,” meaning that ordinary investors are likely experiencing greater market stability and efficiency as a result of these new high-tech trading mechanisms.
What’s the takeaway? In the case of HFT, the data tells a clear and compelling story: electronic trading has improved prices, spreads, access, liquidity and fairness for market participants.
