If you haven’t heard yet, 35 year old James Holzhauer has been killing it on Jeopardy of late. Per fivethirtyeight.com:
Holzhauer has played this game like no one has ever played it before — large bets coupled with expert navigation of the game board. He has now [won 19] games with his total winnings sitting above $1,000,000 and counting, and he is well on his way to surpassing the $2,520,700 won by the most famous “Jeopardy!” record-holder of all, Ken Jennings. One difference? It took Jennings 74 straight victorious shows to bring in that haul, and if he maintains his current pace, Holzhauer is on track to break that record in as few as 34.
Why do we care? Beyond James being from nearby Naperville, IL and a graduate of University of Illinois, which we do some work with through the RCM University Program, it piqued our interest because how he’s doing it is rather quant-like. He goes for high values first, and then attacks the board looking for daily doubles with which to double the gains already gotten. He’s applying his own quantitative model of sorts, albeit run on the wetware of his brain instead of modern computer hardware, on how to go about placing bets and choosing squares on the board.
Where have we seen a quantitative approach to something which had until then been considered mostly an art form…. Oh yeah, trade execution. In the old days, you had a floor broker who did a good job of getting big orders done….in what most considered to be more of an art than a science (although discussions with floor broker reveals similar wetware algorithms at work back then in a back-of-the-envelope fashion). But then along came electronic trading and views of order books and APIs and all the rest. One look at fivethirtyeight.com’s graphic showing where the daily doubles have historically been sure reminded us a lot of something we work with day in and day out.
We’re talking about volume profiles, of course. While a quick glance at that table could be forgiven for seeming like a new type of view for the DOM (or depth of market), with each value representing the number of contracts to be bought or sold at different levels, it can also be viewed as volume buckets across time. Think low value clues ≈ illiquid periods, high value clues ≈ day session hours for a product, and the heavy concentration in the 4th row similar to settlement or closing periods. Just like Mr. Holzhauer likely uses this historical information to inform his algorithm on where to hunt for daily doubles once funded sufficiently, many execution algorithms will analyze a futures contract history manage order placement based on historical models of when volume is likely to occur and help blend order flow into the overall market.
Of course, just like in Jeopardy where you still have to be knowledgeable on the category and get the specific clue correct, knowing when and where to place child orders is only half the battle – you still have to have choose the product, decide the direction, etc. But just like future Jeopardy stars are no doubt analyzing the successful strategies employed by Holzhauer, enlisting a quantitative execution algorithm has become a new best practice.