Session Title

Session 4-2-D: Online Gambling Behavior

Presentation Type

Event

Location

The Mirage Hotel & Casino, Las Vegas, Nevada

Start Date

10-6-2016 10:30 AM

End Date

10-6-2016 12:00 PM

Disciplines

Economics | Finance and Financial Management | Management Sciences and Quantitative Methods | Statistics and Probability

Abstract

In gambling pools, entrants submit predictions and the prizes are awarded to the prediction or predictions closest to actual outcomes. Some well-known examples are football pools (both the global and American game versions), toto, NCAA March Madness bracket pools and horse racing tournaments. For small pools with complete information about outcome probabilities, exact game theory optimal solutions are straightforward to compute. If there is also complete information about the number and strategy of other players, optimal exploitive strategies are even easier to derive. These problems have been treated in the literature.

This paper argues that the complete information approaches are not good approximations to practical optimal strategies, moreover they are usually computationally infeasible. A robust and computationally simple alternative is introduced.

Keywords

Pools, Toto, Brackets, Tournaments, Prediction

Comments

Attachment: PDF containing 19 slides

Share

COinS
 
Jun 10th, 10:30 AM Jun 10th, 12:00 PM

Optimal Strategy for Gambling Pools

The Mirage Hotel & Casino, Las Vegas, Nevada

In gambling pools, entrants submit predictions and the prizes are awarded to the prediction or predictions closest to actual outcomes. Some well-known examples are football pools (both the global and American game versions), toto, NCAA March Madness bracket pools and horse racing tournaments. For small pools with complete information about outcome probabilities, exact game theory optimal solutions are straightforward to compute. If there is also complete information about the number and strategy of other players, optimal exploitive strategies are even easier to derive. These problems have been treated in the literature.

This paper argues that the complete information approaches are not good approximations to practical optimal strategies, moreover they are usually computationally infeasible. A robust and computationally simple alternative is introduced.