Beyond the tote board: A handicapper's algorithm
As a teenager two passions: Horse racing and computers, would form the basis for a future business. My father worked in real estate, and his flexible schedule allowed him to make regular trip to Ellis Park in Henderson, Kentucky. My job was to read him the racing form from the passenger seat. It was in those formative years that the dream of writing a computer program to handicap horses was first born. At the time, handicapping required a stubby pencil and a racing form, but it didn’t take too much imagination for a nerdy kid who could program on a Tandy TRS-80 to get a big idea.
Almost 20 years later, working on my MBA at William & Mary, the dream became a reality. The final project for a Decision Support Systems class, was an algorithm that would become the engine for a horse race handicapping business including website E-Ponies.com, its affiliates and many mobile apps.
The algorithm earned an A in the class. And it worked so well, that I put it online to see if people would pay for the computer-picked horses. At the time the dot.com explosion was reaching its peak, so it certainly seemed feasible. A small business was born.
As it turns out, people would pay for the information because they liked the unique dispassionate analysis that a computer can provide and the clean, easy-to-read format. People paid for the information because it produced a high number of winners. The computer picked horses better than most humans. Over the years, the computer picks have been compared head-to-head against human experts over the course of an entire meet, at Arlington Park and Saratoga Race Course. It consistently performed well against expert handicappers.
The program handicaps the same way as most serious horse players, focusing on three major components to compare a set of horses in a race: Speed, class and form.
- Speed: Not just overall speed, but specifically gate speed – the ability to get to the front early in a race
- Class: An evaluation of the quality of horses with which the horse has been running
- Form: The ability for the horse to finish in the money (“hit the board”)
In evaluating speed, the computer model looks at several factors in the horse’s racing history:
- Where the horse is in the pack after the first quarter mile
- Where the horse is in the pack after the first half mile
- Speed rating: There are several types of speed rating available from different providers but all attempt to normalize the overall speed of a race compared to other similar races run at all tracks. Think of speed rating like an SAT score for races of those specific conditions.
Class is important to a handicapper because class can provide context to other observations. For example, if a horse has been finishing races fifth, sixth or worse, but has been doing so against a top tier of horses, if that same horse drops down to run with cheaper horses, he should perform much better. In my experience class is the single most important factor in evaluating horses in a race. In evaluating class, the computer model looks at several class factors in the horse’s racing history including:
- Position of finish
- Purse sizes
- Earnings per start for the year
- Earnings per start for lifetime