On average, higher ranked players had the upper hand in the Auckland and Australian Open this season. Based on the average winner ranking, Quito and Sofia tournaments featured a lower level of competition (in terms of player ranking).
Older players still outplayed younger competition on average in the Australian Open (+0.36), Auckland (+1.49), and Pune (+1.86) tournaments. The opposite was true at Sydney and Brisbane where the winning player was 1.2 years younger than the losing player.
A random forest model was built using winning and losing players’ rank, height, and age as well as the surface (hard, grass, clay) that the game was played on. The model is trained on 330 matches and tested on 129 matches from the 2018 tennis season (also including Davis Cup matches).
The random forest model suggests that the winner rank, surface, and loser rank are the most important variables for predicting the game outcome. The loser’s height along with the loser and winner age had less predictive performance in the model. Overall, the model correctly predicted 74.4% of the matches in our dataset.
Github: https://github.com/Fremont28/tennis_win_probability-
sportsbrain 