Data show Flightline is best Breeders' Cup horse, but bad bet

Like everyone else, I downloaded the Breeders’ Cup past performances the moment they were available Monday night. My urgency was driven by a desire to see by how large a margin my prediction model, Equus, would favor Flightline in the Classic.

Well, the model has spoken, and its opinions of three of Saturday’s dirt matchups come with some surprises worth discussing. Before diving in, however, here is some context about this analysis. Equus is a machine learning prediction model trained on more than 200,000 individual horse efforts spanning fall 2020 to late september 2022. The current build of the model’s performance was validated against races between Sept. 25, 2022, and Oct. 30, 2022. On this validation dataset, the model’s estimates of post-time odds were more accurate than the morning-line odds. Moreover, among Grade 1 races on the dirt, the model’s favored horse (the horse having the lowest predicted fair odds. i.e. the most likely winner) won 83 percent of those races. Finally, the Equus model computes a measure of positive expected value for each horse in the race, which is derived from estimates of each horse’s win probability in relation to the likely public perception of the win probability. Values greater than 1.0 indicate positive expected value (EV).  Values less than 1.0 indicate negative EV. 

In the Dirt Mile, the Equus model suggests that the morning-line first and second choice (Cody’s Wish and Cyberknife, respectively) will exhibit a neutral value, making this an excellent race in which to single a low-price horse without damaging the EV of your horizontal wagers. This opinion would be best used in the later legs of early horizontal wagers for which the pricing information (such as the tote board, will pays, and probables) are not available to the general public. The model also suggests that the No. 8 horse, Informative, has big upside.

Race 9: Breeders’ Cup Distaff

Plain and simple, the Equus model is fading Nest. This scenario provides a rare opportunity to get positive EV on two other well-known, low-priced horses: No. 1 Malathaat (which the model has as a co-most-likely winner) and No. 4 Clairiere. Secret Oath offers some strong value (+1.9) at a bigger price, but in this spot, the risk of this higher price might not be necessary as Clairiere has an attractive mix of +EV and low odds.

Race 11: Breeders’ Cup Classic

Emotionally, this model prediction was very difficult to accept. I’m driving to Kentucky from Arkansas to see Flightline in the flesh.

All of the replays I’ve watched of this horse, his speed figures, his demonstrated ability to overcome adversity and to run well at the classic distance – all of this suggests to me that Flightline’s true chance of winning should be 90 percent (on the order of 1-10). Only an illness, injury or accident should prevent his trip note from reading something akin to “Cleared. Extended. Much the best.”

But the Equus model is not impressed. Flightline is the most likely winner, but at only 9-5 fair odds there is no value in this horse. My model projects the public might look to Hot Rod Charlie to find value and will bet him down to about 10-1 from his 15-1 morning line. Taiba is the play of the card, offering strong value (2.05) for relatively low odds. In addition, the model indicates that Rich Strike again play might the part of the spoiler (2.03 value at about 9-1 fair odds). My emotional brain will place a $2 win bet on Flightline as a souvenir, but Equus says the smart money is on the rail.