Skip Navigation | Access keys | Accessibility

Horse RacingPokerFootballCasinoRugby Union
Home | Latest Odds | Top Offers | Bookmakers & Casinos | Responsible Gambling Log In | Register

Contents

Pages

Home > Books > Forecasting methods for horseracing

Book Extracts

Forecasting Methods for Horseracing

Author: Peter May
Publisher: High Stakes
Buy this book at Amazon

<< Previous Page   Next Page >>

Forecasting Methods For Horseracing

Whichever approach you choose, the horseracing problem offers a challenge worthy of our best efforts.

____________ The Horseracing Problem

Thoroughbred horseracing in Great Britain has been extremely well documented for many years. Records still exist detailing the very first organised horseraces, such as the Newmarket Town Plate in which King Charles II was successful in 1671. Consequently, a wealth of information is available to the race analyst, concerning all races and the horses that compete. The historical records include the peculiarities of each race track, whether right or left handed, undulating or flat. For each horse, all previous race details including times, race distances, course conditions and race commentary, together with the animal's pedigree are documented. Jockey and trainer statistics are also available indicating success rates by several variables including track and race type. The fact that racing is so well documented is helpful to the race analyst by providing the basic information with which to work.

Although the availability of this large volume of data is advantageous for modelling purposes, the level of information relating to each horse in a race is extremely detailed which severely complicates the task of generating workable systems. Discussion of race analysis methods with recognised racing experts suggests that this level of detail, coupled with a lack of structured approaches to race analysis, has resulted in wide disagreement between the experts regarding optimal solution methods. Although there is general agreement between the experts in the identification of horses with either very high, or very low, probabilities of success, there is considerable disagreement for less well-defined runners. Interestingly, these problems mirror those found when developing computerised methods for other prediction problems, such as assessing mortgage applications2.

A second problem with the horseracing domain is the competition between the runners in a single race. A horse may possess outstanding winning credentials, but the likelihood of its success is also dependent

11