Levels of the Game: A Statistical and Mathematical Analysis of ATP Grand Slam Competitions from 2005-2012

Abstract

Data analytics is one of the most widely used processes around the globe to help solve problems and optimize results. This process, however, has been completely under-utilized within the global tennis community, a community that is ripe with data but lacking in analysis. This thesis contributes to the growing movement of data analysis in professional tennis by examining men’s Grand Slam point data from 2005-2012. Using basic statistical analysis and binary logistic regression modeling, this work investigates four major topics in professional tennis that have been discussed on a variety of levels for years, but lack statistically-backed claims. The four topics investigated are: serving and volleying, the three di↵erent court surfaces, one-handed and two-handed backhands, and five set matches. The analysis concludes that though serve and volleying is decreasing in use, it is not becoming less beneficial to use, and in fact should be utilized more often to increase players’ winning probabilities. The work also finds that clay court points are not longer than hard court points, and that first serve percentage is most important on grass courts. For backhands, the results find that though the use of one-handed backhands is decreasing, one-handed players are producing higher winner percentages and lower error percentages than two-handed players. Lastly, the analysis identifies service points won as the key variable separating final set winners from losers. Overall, the results confirm the idea that increased analytics can impact strategy at the highest level of tennis, while also signifying the potential change analytics can make to the overall game in the near future.