ARTIFICIAL INTELLIGENCE FOR HANDICAPPING

Recently, I received an email pertaining to an artificial intelligence program for handicapping horse races.  The vendor charges a monthly fee of $169.  Also, this past week I saw artificial intelligence predictions for who would win U. S. Open tennis tournament matches and for college and professional football games.

Artificial intelligence models are based on machine learning.  Vast amounts of historical data are analyzed via computer programs to produce insights and predict outcomes.  For example, IBM’s Watson used over 7 million data points to predict the probable winners of every match in the U. S. Open.  The data included player past performances and what media and experts are saying positively or negatively about each player.

About a year ago, I subscribed to an artificial intelligence model for buying and selling stocks.  Various models or “kits” were available, depending on an investor’s objective.  To illustrate, four of the kits were Value, Emerging Tech, Short Squeeze, and Inflation Protection. The model I selected was Capital Preservation.  Naturally, some of the kits performed better than others in the market volatility of summer 2022.  Most underperformed the S & P 500 and the Dow Jones Industrials. That was to be expected if a model’s goal was out of sync with prevailing market conditions.

Each of the stock buying and selling kits was based on analysis of its own set of historical data.  Bitcoin Breakout and Large Cap, for example, obviously were modeled using different data.

This brings me to my major point about artificial intelligence in horse-race handicapping.  A generic model developed with data from a potpourri of racetracks and conditions would not have the predictive capability of a “kit” developed for, say, 6 furlong races at Belmont Park.

While I am a proponent of the ethical use of artificial intelligence in all kinds of human endeavors, the caveat is that a model is only as accurate as the data fed into a machine learning program. Case in point: A machine learning model crafted with data from European turf races would be highly suspect for handicapping American dirt races.

Copyright © 2022 Horse Racing Business