17th Annual ECSS-Congress, Bruges 2012

Abstract details

Abstract-ID: 1390
Session: [MO-PM32] Exercise, Nutrition & Metabolism
Lecture room: Auditorium
Date & time: 03.07.2014 / 15:00 - 16:00
Authors: Viney, M., Bedford, A., Kondo, E.
Department: School of Mathematical and Geospatial Sciences
Country: Australia
Abstract text Introduction When a tennis match is in-play, the viewer has knowledge on in-play serving statistics such as first and second percentage of serves in and won. There has been various works involved in applying serving statistics to increase serving performance and to find the optimal serving strategy. Barnett et al. (2008) suggest separating serving statistics in relations to the surface of the court, Gale (1971) applied a simple mathematical model, whilst George (1973) applied a probabilistic model. Although various research has been performed, limited research has involved in applying in-play serving statistics to increase serving and player performance. The aim of this study was to apply in-play serving statistics into a set simulator to determine whether this application is an effective coaching tool for player improvement. Methods The set simulator allows the inclusion of faults to increase the accuracy of the probability of winning a set in tennis. This simulator has the ability to adjust the serverís probability of winning a point on first and second serve at any phase of the set in order to calculate the likelihood of each player winning the set. To test the effectiveness of the set simulated, it was compared against the Markov Chain model that was applied to a case study. The Markov Chain model is generally used in the tennis literature to calculate the probability of winning a game, set and match before and during a match. To increase the accuracy of the results, serving statistics are collected from the first set and applied to calculate the probability of winning the second set. Results Results showed that the simulator was superior in estimating the outcome of the set. In comparing the two methods, the Markov model gave an extra 0.13 probability to the underdog to win the set, where the favourite won the set 6-1. Thus the simulator was more effective in this case study. By increasing the underdogís serving statistics by two percent, resulted in increasing the simulated probability of winning the set by six percent. If the underdog decides to change serving strategy by increasing the percentage of first serves in by five percent, the simulated increased the likelihood of winning the set by three percent. Discussion The results suggest that the simulator that incorporates the faults and in-play match statistics is a more effective approach than using the Markov Chain model. From the results showed, this simulator can be an effective coaching tool to improve playerís performance such as altering serving strategy to improve the outcome of the match. References Barnett T, Meyer D, Pollard G. (2008). Med Sci Tennis, 13(2), 24-27. Gale D (1971). Mathematics Magazine, 44, 197-99. George S (1973). Appl Stat, 22, 97-104.
Topic: Sport Statistics and Analyses
Keyword I: In play
Keyword II: tennis
Keyword III: coaching