[CatfishRivers]

What do you'll think about the Universal Tennis Rating (UTR)? Tennis Channel, Federer/Djokovic and many others endorse it. Here's a definition of the UTR:

** Universal Tennis Rating (UTR) is a reliable, accurate index that rates players' skill in tennis. UTR rates all players—men, women, and children—on a single 16-point scale (with two decimal places, e.g., 11.29) that works for players globally regardless of their skill level, from beginners to top professional competitors.

UTR's algorithm calculates ratings from the last 30 eligible matches played within the preceding 12 months. The main data points are the percentage of games won (not simply the won/lost result), and the opponent’s rating. Each match played generates a rating; a player’s UTR is a weighted average of all eligible matches. **

https://en.wikipedia.org/wiki/Universal_Tennis_Rating
https://www.myutr.com/

Here's the top 10 WTA players going by current UTR:

1) Andreescu - 13.43
2) S. Williams - 13.36
3) Barty - 13.35
3) Osaka - 13.35
5) Pliskova - 13.32
6) Kenin - 13.28
6) Kvitova - 13.28
8) Keys - 13.27
8) Svitolina - 13.27
10) Konta - 13.26
10) Halep - 13.26

*Note: There were several players tied with the same UTR. In those cases with ties I listed the player with the higher trending 3 month UTR first.

 

[amirbachar]

UTR is not meant to rank the top players accurately, it is meant to have a relative ratings for millions of players.

 

[skytennis]

Yes, it is a very good rating. Much more accurate than the real rankings if you want to see who is the better tennis player

Sent from my SM-N920C using Tapatalk

 

[Darop.]

Not a bad idea, an even more specific version of quality points in a sense.


What really is a "make it or break it" issue is how much weight to give to ranking difference between players, what coefficient to use based on opponent's ranking.


I really don't get one thing though: if the system is based on a) % of games won and b) opponent's ranking, how do you actually start using this ranking system if you can't calculate a players' score without knowing the score of the opponents defeated? :shrug: Seems like a kind of chicken-egg dilemma

 

[CatfishRivers]

The top 10 UTR ratings for WTA players all seem very close. Just fractions of a decimal point separate them. Which brings me to the question: Does anybody know how accurate the UTR is in predicting the winner of a WTA match? On the UTR web site they also give each player a "trending 3 month UTR". Would that be more useful in predicting a winner?

 

[amirbachar]

Not a bad idea, an even more specific version of quality points in a sense.


What really is a "make it or break it" issue is how much weight to give to ranking difference between players, what coefficient to use based on opponent's ranking.


I really don't get one thing though: if the system is based on a) % of games won and b) opponent's ranking, how do you actually start using this ranking system if you can't calculate a players' score without knowing the score of the opponents defeated? <img src="https://www.tennisforum.com/images/smilies/shrug.gif" border="0" alt="" title="Shrug" class="inlineimg" /> Seems like a kind of chicken-egg dilemma

It's a mathematical object called Makrov stationary distribution. Here is a relatively simple explanation:
https://brilliant.org/wiki/stationary-distributions/
The process of computing it is starting from random ratings and then each iteration converage to the real ratings.

 

[Darop.]

It's a mathematical object called Makrov stationary distribution. Here is a relatively simple explanation:
https://brilliant.org/wiki/stationary-distributions/
The process of computing it is starting from random ratings and then each iteration converage to the real ratings.

Oh cool :inlove:

 

[Ceze]

The top 10 UTR ratings for WTA players all seem very close. Just fractions of a decimal point separate them. Which brings me to the question: Does anybody know how accurate the UTR is in predicting the winner of a WTA match? On the UTR web site they also give each player a "trending 3 month UTR". Would that be more useful in predicting a winner?

yes it's quite tight, for example today Mladenovic UTR 12.93 (UTR#59) def. Barty UTR 13.35 (UTR#3). Many players between 13.35 and 12.93 (56 players) :SNE:

 

[goldenlox]

I think its good for weekend hackers and regular club players to find an opponent their level
But 3 US players in the top 8 makes it sound dopey. There were no US players in Shenzhen but the 2nd alternate