BETTING ON ARTIFICIAL INTELLIGENCE

Analyzing and betting sensibly on horse races is an intellectual challenge requiring consideration of a multitude of variables that may affect the outcome.  A handicapper must sort out the variables and weight them in terms of their importance given the conditions of a race.  Computer software enables a handicapper to develop a quantitative model and back-test its predictive power.

Might artificial intelligence produce a superior result?  In 1997, IBM’s Deep Blue defeated chess champion Garry Kasparov.  Great advancements in AI have been made since then, in a variety of dissimilar endeavors, from medicine to gaming.  For example:

  • Northwestern University researchers developed AI that beat the average American in solving the intelligence test of visual and analogical reasoning called the Raven Progressive Matrices Test.
  • Researchers at Imperial College London developed AI that was more accurate in diagnosing pulmonary hypertension than the typical cardiologist.
  • AI is being applied by major employers to sort out the best candidates to fill job openings.
  • At the Rivers Casino in Pittsburgh, Pennsylvania, researchers from Carnegie Mellon University used the AI program Libratus to play four world champion poker players (one-on-one, not as a group) in 120,000 hands of heads-up no-limit hold’em.  In the end, Libratus had an advantage of 1.8 million chips.

There are important fundamental differences, of course, in playing chess and poker, and handicapping horse racing or sports events.  In chess, both players can see the other’s moves and thus there is no private information.  By contrast, in poker, each player has incomplete information because some of the cards are hidden…and thus probability becomes a key determinant of winning and losing.  In horse racing and human sports competition like soccer and basketball, there is plenty of public information available to bettors, but also there is private information that the average bettor is not privy to…and luck plays a role as well.

AI may be more difficult to develop for handicapping horse races than playing chess or poker, but it does offer potential because AI can handle the interactions among numerous variables that determine the results of a race and it can do simulations.  Plus, AI takes human emotion out of the decision.

Our sponsor, SBAT, has been exploring possibilities of using AI to better understand, interpret and explore statistics in an effort to make betting easier and offer valuable information and guidance needed to wager intelligently on sports like horse racing and soccer.

Copyright © 2017 Horse Racing Business