Strategy optimization in beachvolleyball : applying a two scale approach to the olympic games

Titelangaben

Hoffmeister, Susanne ; Rambau, Jörg:
Strategy optimization in beachvolleyball : applying a two scale approach to the olympic games.
2017
Veranstaltung: MathSport International 2017 Conference , 26.-28.06.2017 , Padua (Italy).
(Veranstaltungsbeitrag: Kongress/Konferenz/Symposium/Tagung , Vortrag )

Weitere URLs

Abstract

Our two scale approach uses two connected Markov Decision Problems (MDPs) to give strategy recommendations for the olympic final in beachvolleyball. The approach relies only on the individual player skills that are estimated from analyzing video records of the pre final matches. In the more detailed MDP, called gameplay MDP, these skills are used to simulate the outcome of serves and field attacks for a certain pairing of teams. Nearly every thinkable team strategy can be implemented and simulated. After knowing the probability for a direct point and fault after a serve or field attack, it is possible to determine the strategy with the highest winning probability. For this purpose a second, more aggregated MDP is used. When using perfect knowlegde about the final match, that means, we use a strategy that is similar to the strategy in the final match and skills estimated from the final match, we can estimate the outcome of serves and field attacks with an average absolute error of three percentage points. Furthermore, we derived analytical bounds for the simulated probabilities in terms of the individual player skills. They are evaluated and compared with the simulation results of the men’s olympic final. In order to give an overview of the optimal strategy for varying opponents performance and the own form on the day, we invented skill-strategy-score cards. They tell at a glance how sensitive the optimal strategy is.