Trail Of Cookie Crumbs

A recent paper titled “The Volume Clock: Insights Into The High Frequency Paradigm”  takes a very different approach to analyzing the effects of HFT.  The authors (Easley, Lopez de Prado and O’Hara) concede that even without their speed advantage, HFT would still be able to exploit institutional and retail orders.  They note that HFT takes advantage of market microstructure factors that have been created due to regulations like Reg NMS and MiFID.  They claim that the Efficient Market Hypothesis is no longer valid in an HFT world and that over short intervals, “prices are predictable artifacts of the market microstructure“.  The authors state that “HFT reacts to information leaked by LFT (low frequency traders) in order to anticipate their actions“.   In other words, institutional and retail orders are leaving a trail of cookie crumbs that HFT’s are easily sniffing out and using to their advantage.

Indeed, not only does this paper point out that HFT’s take advantage of the predictable actions of institutions and retail, but they claim that predatory HFT can actually cause these events to occur:

“Rather than possessing exogenous information yet to be incorporated in the market price, they know that their endogenous actions are likely to trigger a microstructure mechanism, with foreseeable outcome. Their advent has transformed liquidity provision into a tactical game.

Examples of this behavior are quote stuffing, quote danglers, liquidity squeezers and pack hunters.

While highlighting these dangerous predatory HFT behaviors, the authors do seem to give a pass to the HFT automated market makers.  They claim that these predatory activities even harm HFT market makers and that HFT market makers “sometimes may suddenly pull all orders, liquidate their positions and stop providing liquidity altogether.”  We think this is a major problem.  Our market has replaced traditional market makers that had quoting and trading obligations with HFT market makers that can now suddenly pull their quote at the sign of trouble.  This is the real problem with our markets but let’s get back to the paper for now.

The authors also highlight an example of predictable behavior that HFT’s take advantage of:

There is no question that the goal of many HFT strategies is to profit from LFTs mistakes.  We have taken a sample of E-mini S&P500 futures trades between 11/07/2010 and 11/07/2011. We have divided the day in 24 hours (y-axis), and for every hour, added the volume traded at each second (x-axis), irrespective of the minute. For example, E-mini S&P500 futures trades that occur at 20:20:01 GMT and 20:23:01 GMT are added together.10 This analysis allows us to see the distribution of volume within each minute as the day passes, and search for LFTs executing their massive trades on a chronological time- space. The largest concentrations of volume within a minute tend to occur during the first few seconds, for almost every hour of the day. This is particularly true at 02:00-03:00 GMT (around the open of European equities), 13:00-14:00 GMT (around the open of U.S. equities) and 20:00- 21:00 GMT (around the close of U.S. equities). This is the result of TWAP algorithms and VWAP algorithms that trade on 1 minute slots. A mildly sophisticated HFT algorithm will evaluate the order imbalance at the beginning of every minute, and realize that this is a persistent component, thus front-running VWAPs and TWAPs while they still have to execute the largest part of the trade.”

Pretty simple stuff that even a “mildly” sophisticated HFT algo can detect.  Good luck human!