Institutional Algo Predictability

Yesterday we came across an interesting blog post by John Cochrane, a professor at the University of Chicago’s Booth School of Business. His post, titled Weird Stuff in High Frequency Markets, highlighted the following graph from a 2011 academic paper by Hasbrouck and Saar, titled Low Latency Trading:

This chart is interesting, as it shows the flow of messages on NASDAQ over any typical 10-second interval. Note that there are large quote traffic spikes every second, on the second exactly, followed by a second smaller surge in traffic on exactly the half-second. Professor Cochrane explains:

Evidently, lots of computer programs reach out and look at the markets once per second, or once per half second. The programs clocks are tightly synchronized to the exchange’s clock, so if you program a computer “go look once per second,” it’s likely to go look exactly on the second (or half second). The result is a flurry of activity on the even second.

He goes on to state:

It’s likely the even-second traders are what Joel and Gideon call “Agency traders.” They’re trying to buy or sell a given quantity, but spread it out to avoid price impact. Their on-the-second activity spawns a flurry of responses from the high frequency traders, whose computers monitor markets constantly.

And this brings us to what is on our minds. We all use algorithms to execute at times, whether they be VWAP oriented, POV (percentage of volume) oriented, or more sophisticated in nature. And all of these algorithms shred our orders into smaller bits and pieces and deliver them to the market. So many of our notes to you have focused on the routes they take as they are delivered for execution. And so many of our notes to you have stressed the importance of avoiding predictability.

Professor Cochrane’s blog post reminds us that we also need to focus on the timing of the delivery as well. His blog post, and the chart we included above, illustrate that our algos are indeed all synced to stock exchange clocks, and in general deliver our shredded order flow to the markets at predictable times – on the top of the second for example. But as we all know, HFTs are operating literally millions of times faster than that (microseconds), and are cancelling, pinging, quoting, and cancelling in loops. If Hasbrouck can see the pattern in the chart above, and Professor Cochrane can, and all of us on the investment buyside can, then you can bet a Bernanke Billion that predatory HFTs can.

So. What to do? I think it behooves all of us to start asking all our algorithmic providers for more detail on how our orders are delivered to the market, including the timing. I think a little randomness and unpredictability in the delivery might go a long way towards helping us all improve our market impact.