What I learned from betting the Kentucky Derby -- and why I keep doing it, happily, even when I lose

My best Kentucky Derby came in 1987, when I pooled my bets with a widening circle of friends who were convinced that my ability to divine the future from the arcane data of the Daly Racing Form’s past performances was the ticket to Derby Day bragging rights and modest riches. I picked Alysheba to win, and Bet Twice to place – a result that seem ordained by my acumen when they crossed the finish line in that exact order. I still remember the pari-mutuel clerk peeling off hundred dollar bills when the results were posted, looking up at me through the window, and saying “Nice lick, kid.” Yes! I had arrived as a horseplayer.

Forget March Madness brackets or football points spread. The Kentucky Derby is the premier intellectual challenge for sports bettors, one which continues to teach us about our cognitive biases and humble our pretensions to wisdom.

For those who want to know my picks for the 2017 Derby, go to the end of this post. For those who want more context and a reason not to go overboard in your betting, please keep reading.

I have learned a few lessons and earned a bit of modesty since 1987, when Alysheba’s win made me feel like an apex predator amidst the crowds of occasional gamblers who juice the betting pools of Derby Day. Just as we have gained new understanding of human cognition and decision making since then -- think of the work of Daniel Kahneman and Amos Tversky profiled in Michael Lewis’ The Undoing Project, I and my fellow thoroughbred handicappers have since experienced a hard-earned maturation of our ability to manage data, wrestle with the known unknowns of racing results and the known knowns of pari-mutuel odds making.

When I first started betting the Derby, Big Data was in its infancy, and formulas for predicting Derby winners were spun from the recurrence of patterns which Thomas Bayes, an 18th Century pioneer of statistics who developed what he called “the doctrine of chances,” would have recognized and appreciated. We start with working models of the world and adjust them as they fail. Thus, by the time Secretariat won the Derby in 1973, handicappers divined that the keys to victory were to be found in a trifecta of past performance profiles that included having run as two-year old, having won a graded stakes race and having a certain “dosage” profile based on the stamina of one’s sire and dam. Although it seems like magical thinking now, I was happy to invoke this model of the world to cash some large winning tickets on Derby winners like Gato del Sol (1982), Ferdinand (1986) and the Alysheba-Bet Twice exacta in 1987.  But then winners like Winning Colors, Go for Gin and Grindstone confounded the models. You don’t read much about dosage indexes any longer. Reality forces us to modify old theories or discard them.

  Enter the quants. By the 1990s, Andrew Beyer and others had developed and refined systems that quantified horses’ running times by adjusting for the effects of different racing surfaces and the “shape” of a race that might favor a front runner or a closer. Today “the Beyer figure” has become a commonly-used metric for comparing the speed of horses competing on different racetracks. This is important in the Derby, as in March Madness, since it is the first time that a new crop of athletes who have proved their strength in different regions get to compete in same location.

  But horses with “the best Beyer” rarely win the Derby, which has forced the quants to chase more and better data to predict racing outcomes, in much the way that Bayes described how we forever adjust our models of the world to our experience of it.

  Meanwhile, the old hands who like to assess the subjective qualities of class, competitiveness or the condition of a horse continue to struggle with the changes in training techniques, inflated purses in foreign venues like the United Arab Emirates and the sequences of races to the Derby, creating new Bayesian models, touting them when they win, and refining them when they lose.  

  But picking winners is one thing, making money is another – and this is where I mourn the loss of regional betting pools. I used to place my Derby bets through friends and relatives in different parts of the country in order to take advantage of the biases that a hometown crowd will demonstrate in betting a local favorite. As Kahneman and Tversky have documented, we are all prone to motivated reasoning, even when we think we’re being objective. But now that technology has allowed the rapid consolidation of betting totals into one national betting pool, the potential to exploit local biases has disappeared.

  Still, the Derby is an event that continues to attract the opposite of smart money. So for those of us who think we can outsmart the next guy and are less prone to the biases that affect others’ thinking, Derby 2017 offers another chance to choose hubris over humility. If you win, you’ll feel smart. If you lose, you’ll come up with some plausible explanation for why your horse woulda-coulda-shoulda done better. In this spirit, here is how I’m approaching Kentucky Derby 2017.

  As always, I’m looking for horses to break the mold of old models.

  First on that list is Thunder Snow. No foreign horse shipped from overseas has ever won the Derby. Thunder Snow was bred in Ireland and his last race was run in Dubai. He is my Bayesian pick of the race. If he wins, we will have to redo all our models, kind of like we are doing now in electoral politics.

  Second is Hence, who sports a respectable Beyer speed rating, good looks and a less respectable venue (New Mexico) for his last win. With the exception of one race at Saratoga last year, this horse has been running in flyover country, so he is likely to be overlooked in a national betting pool dominated by the coastal elites.

  Third is Gunnevera, the woulda-coulda-shoulda horse of the Florida Derby. Yeah, he lost that one but coulda done better if he didn’t break badly from an outside post. Plus, his mom is a mare sired by Unbridled, who was my own mom’s favorite Derby winner. That’s a sentimental reason, I admit. But, as Kahneman knows, I’ll find a rational basis for liking him if he ends up winning the race.

  But don’t overlook the favorites – Always Dreaming, Classic Empire and McCraken – if you believe in the wisdom of crowds. Of the three, I’d favor McCraken for my trifectas.

  Finally, perhaps the purest form of handicapping, akin to the horse whispering that has come into vogue of late, is the ability to read a horse’s fitness and eagerness to run on any given day. The bible for this approach is The Body Language of Horses by Bonnie Ledbetter, a classic in its field. So, if you wait to place your bets until after the post parade, look for what Ms. Ledbetter describes as the telltale signs of the “sharp horse” – one that is prancing, on its toes, with a gleaming coat and arched neck, but not fractious or sweaty. There is no mathematical formula for spotting these horses; it’s more art than science. But anyone who picks a winner on this basis will get special bragging rights, for there is no disputing that you are communing with a reality that cannot be reduced to statistics when you pick a winner on looks alone.

  I still remember the elation of my friend Marvin, when he picked Pass Catcher in the 1971 Belmont, because of how fit the horse looked on the track. Pass Catcher won at 34-1. Marvin had his moment, as did I in 1987. A neuroscientist today could probably describe the experience in chemical terms. We just enjoyed it. And we’ve been trying to repeat it ever since.

  Enjoy the race.