"TenisRank": A new ranking of tennis players based on PageRank

TL;DR

This paper introduces 'TenisRank', a novel tennis player ranking system based on PageRank, aiming to better reflect players' chances of winning and improve upon traditional ATP rankings.

Contribution

The paper develops a new PageRank-based tennis ranking using match history data, offering a potentially more accurate prediction of match outcomes.

Findings

The new ranking correlates well with match results.

It provides better predictions of match winners.

It offers insights into player performance and tournament characteristics.

Abstract

In the light of the need to achieve a ranking which is understood by all tennis supporters, the ATP ranking is exposed to constant complaints from players and at the same time exposes new players to be benefited with a good tournament to be able to start progressing in their careers. Moreover, the ATP ranking is not powerful enough to predict with certainty who will be the winner of a match if we are based solely on the positions. In order to combat these problems, the idea of creating a new ranking that can indicate what are the real chances of victory of a player before the start of a new tournament arises. Based on the PageRank method, generated by Larry Page and Sergey Brin, we created a new ranking that specifically uses the characteristics of the tournament to generate data. Based on a history of 40,000 matches, we intend to evaluate how the new method is performed as compared to other existing rankings in order to analyze if we really achieved an improved and real reflection. Once we have obtained the ranking, we intend to evaluate, taking a sample game, the ranking of the players that dispute it and the characteristics of such game to be able to indicate the precise probability for the player with better ranking to win the game.