Show HN: Tennis winning probability analyzer based on Markov model

Show HN: Tennis winning probability analyzer based on Markov model(tenniscalc.pythonanywhere.com)

3 points by vander91 2 years ago | 2 comments

I built a simple tool to compute winning probability on a 3-set tennis match, from any possible score. Just a fun little project I always had in mind

Bostonian 2 years ago |

Interesting, thanks. Here is a similar question I have wondered about. Suppose a player wins say 30% of points when returning. What is his chance of winning a N-set match (N = 3 or 5) as a function of his probability p of winning points on serve? By symmetry we know it is 0.5 for a p = 70% chance of winning service points, but how quickly does it increase/decrease as p varies from 70%? One could do a Monte Carlo simulation and plot the result. We know there is a greater chance the better player wins a 5-set match than a 3-set match, because the longer the match, the less variance there is in the result.

There are probably serious sports bettors who already have a similar app to yours to aid them in betting on tennis matches during the matches.

vander91 2 years ago | |

You can use the above tool to answer. If player1 wins 30% of points returning, that implies that player2 wins 70% of points serving. Input that number for p2 in the calculator, and check how the winning probability changes as you change p1. As you pointed out, at p1=70% the match proba will be 50%. At p1=65%, proba drops to 27.4%

In general, tennis matches are very sensitive to this point winning probability.