HFT and the Movie, The Sting

The 1973 Robert Redford / Paul Newman blockbuster The Sting hit the big screen when I was but seven years old, but I didn’t’ get a chance to actually see the movie for the first time until I was about twelve, when my dad first brought home our first VCR. I remember watching the movie with him, and pointing at the screen when that Lonnegan character appeared (Robert Shaw) on screen. I remember telling my dad, “Whoa! Awesome. It’s Quint from Jaws!”

I also have memories of piano lessons with a stern teacher who was attempting to teach me the soundtrack. Attempting. But I digress…

The Sting tells the story of a well-orchestrated con in which a ruthless gangster, Lonnegan, becomes convinced by Kelly (Redford) that together they can make a risk-free fortune by betting on horse races that have already been run,

Lonnegan: Got a system, Kelly?
Hooker: No. You can still lose with a system.
Lonnegan: You’re past-posting, aren’t you?
Hooker: Could be.
Lonnegan: How?
Hooker: You’re gonna stay in?
Lonnegan: Not until I get some answers.
Hooker: I got a partner downtown. He runs the central office of the Western Union. Race results from all over the country come in…and go across his desk to the bookies. All he does is hold them up until he can call us to get a bet down on the winner. Then he releases the results to the bookies. We clean up on a race that’s already been run.

In the past, we have used the analogy of HFT and betting on races that have already run; see our white paper on Latency Arbitrage. You see, taking advantages of the two speeds in the market (actually there are more than two speeds), opportunistic HFTs armed with tools from the stock exchanges can profit from knowing the real value of a financial asset, like a stock, microseconds ahead of the public. In earlier years HFT and latency arbitrage were the best kept secrets on Wall Street. Today, there are thousands of small HFT firms crowding the “game” and reducing its marginal profitability.

What is worrisome, though, is that in lean times some HFT’s will do anything to maintain their edge. In a TEDx conference last year, Yan Ohayon brought up the topic of “quote stuffing”, in which he discussed how HFTs slow down “the quote” by stuffing data into the market (go to the 7:15 mark). Nanex LLC has also documented numerous “quote stuffing” instances in their research. We have also noted numerous instances of thousands of quote changes/orders bombarding thinly traded stocks for no discernible reason, except perhaps the slowing down of a specific stock exchange server in which the targeted “stuffed” thinly-traded stock happens to be traded alongside of a more active stock, which perhaps is ultimately the real target of the latency-arbitraging HFT algorithm that is producing the quotes.

When we look at the May 6th Flash Crash of 2010, as well as the countless “mini flash crashes” that occur almost daily in the market place, we see how crucial it has become in our market structure to maintain uninterrupted big-data flow. When there is any kind of disruption or speed discrepancy in the flow of this data, spooky things start to happen. The CFTC/SEC Joint Commission examining the May 6th Flash Crash specifically describe the data delays and disruptions on that day, and note their role in the ensuing market-wide liquidity vacuum. The bottom line is that “market makers” disappear in the uncertainty of not being to price their inventory without delay, even if the delays are sub-second.

To us it is amazing that we have become so dependent on the “liquidity provision” of HFTs whose games depend on speed, data timing differences, and … like in the movie The Sting… betting on races that have already been run. I highly doubt that this was the result that regulators were going for when they listened to industry insiders, and crafted the rules that they did over the past 15 years. Perhaps it is time for our regulators to conduct their own “sting”. Perhaps they can realize that what hooked Lonnegan in The Sting was the realization that data-flow disruptions were bad for bookies, but good for him.

By the way, credit for the borrowed train of thought in this note needs to be given to Paul Wallis, a CTO at Obashi, who wrote a blog article about data-flow disruptions, and The Sting, back in June 2011. Thanks Paul!