bei Google ScholarTitelangaben
Hoffmeister, Susanne ; Rambau, Jörg:
Markov Decision Processes for Sport Strategy Optimization.
2018
Veranstaltung: 23rd International Symposium on Mathematical Programming (ISMP)
, 01.-06.07.2018
, Bordeaux (France).
(Veranstaltungsbeitrag: Kongress/Konferenz/Symposium/Tagung
,
Vortrag
)
Abstract
We introduce a framework of Markov Decision Processes suitable for sports games. The objective of these so-called Sport-Strategy Optimization Markov Decision Processes (SSOMDPs) is to maximize the probability of winning a match. We investigate properties of SSOMDPs that help to determine an optimal strategy. The introduced SSOMDPs are used to answer a sport strategic question. We formalize what we consider a sport strategic question and show how the optimal strategy may depend on the involved opponent team and the environmental circumstances of the specified match. The talk includes examples of beachvolleyball-SSOMDPs that model the same strategic question on different levels of detail. We show the advantages and disadvantages of each modeling decision from a sport-related perspective. Finally, we introduce a framework that combines the advantages of a rough model with a very detailed model. We call this framework a two-scale approach.