Recent inflation of UTR ratings?

[Moon Shooter]

Yes if a captain it can be useful… but I wonder if your eye-test is better? For ppl you haven’t seen play UTR or TR or WTN can guide you. But I would think you could predict better than the machine how the lines would do?

Yes I think my eye test or even playing with my players can be a better test. Usually the ratings are in line with my views there but sometimes tennis record in particular can be way off. On my Tri level team it had a 3.0 self rated rated higher then the 4.0 players.

Tr and utr can be helpful to get a sense if a team has a deep bench or what team might be the team to beat at regionals. It can also be helpful to see what competition my players played against compared to how they did. Also if I see two players as about equal I look to tr and utr to break a tie.

I don’t have any faith in wtn to use it yet.

 

[Roforot]

Yes I think my eye test or even playing with my players can be a better test. Usually the ratings are in line with my views there but sometimes tennis record in particular can be way off. On my Tri level team it had a 3.0 self rated rated higher then the 4.0 players.

Tr and utr can be helpful to get a sense if a team has a deep bench or what team might be the team to beat at regionals. It can also be helpful to see what competition my players played against compared to how they did. Also if I see two players as about equal I look to tr and utr to break a tie.

I don’t have any faith in wtn to use it yet.

I'm starting to believe in WTN at least more than UTR.
I don't like what I'm hearing about how WTN is handing out starting #s based on age/gender, but amongst people I'm playing against or seeing in tournaments, it seems to be solid.
UTR for example had a girl at 4.42 (for some reason I could see her 2 decimals) but she's been on a tear in womens' and mixed. Even winning a 4.0 womens tournament as a 3.5 player.
UTR had several people at 5.xx that she beat. WTN and TR both reflected her much more accurately. The argument that she's just beating low ranked player is a non-sequitour. She's beating the opponents ahead of her, many of whom are according to UTR rated above her!

 

[travlerajm]

I'm starting to believe in WTN at least more than UTR.
I don't like what I'm hearing about how WTN is handing out starting #s based on age/gender, but amongst people I'm playing against or seeing in tournaments, it seems to be solid.
UTR for example had a girl at 4.42 (for some reason I could see her 2 decimals) but she's been on a tear in womens' and mixed. Even winning a 4.0 womens tournament as a 3.5 player.
UTR had several people at 5.xx that she beat. WTN and TR both reflected her much more accurately. The argument that she's just beating low ranked player is a non-sequitour. She's beating the opponents ahead of her, many of whom are according to UTR rated above her!

WTN is weak on the self-rating with over-use of pre-assumptions (like the age penalty), but gains momentum and is probably the most accurate of the rating systems for players that have a substantial amount of data in the system. It quickly recovers accuracy if you log matches with opponents of different ages.

UTR is pretty solid for singles when players are active and playing a lot of matches. But outside of that UTR kills its itself with bad assumptions. By continuing to adjust rating after a player goes inactive, it compromises accuracy of whole system and goes unstable. And the stupid 2-unit gap rule in the algo, along with a bug in the algo for how it calculates opponent strength and partner strength adjustment in doubles, makes it useless for mixed doubles, ignoring a large fraction of the only matches with both genders on same court at same time. It overweights opponent strength, so it is more of a rating of your strength of schedule than it is of the player, giving illusion of accuracy. That’s how you end up with stupid results like Jenson Brooksby briefly rising to UTR World #1 this summer.

TR is pretty solid for rec level play in both singles and doubles, with bonus for separating out mixed. Algo is simple and logical and easy to reverse engineer. It doesn’t separate singles and doubles rating, so can be misleading if you don’t review the match history to estimate those ratings separately yourself. It’s also a bit less sticky than the other rating systems as it only averages last several match ratings, so it has more recency bias and more sensitive to outlier results.

 

[Moon Shooter]

I'm starting to believe in WTN at least more than UTR.
I don't like what I'm hearing about how WTN is handing out starting #s based on age/gender, but amongst people I'm playing against or seeing in tournaments, it seems to be solid.
UTR for example had a girl at 4.42 (for some reason I could see her 2 decimals) but she's been on a tear in womens' and mixed. Even winning a 4.0 womens tournament as a 3.5 player.
UTR had several people at 5.xx that she beat. WTN and TR both reflected her much more accurately. The argument that she's just beating low ranked player is a non-sequitour. She's beating the opponents ahead of her, many of whom are according to UTR rated above her!

UTR complety seperates out sungles and doubles. Because usta does not have many singles matches there can be quite a few rating islands. So yes the singles can often be off.

A utr 4.42 is somone playing at mid to upper 4.0 for womens tennis. If she was a 3.5 it sounds like ntrp had her mis-rated as well. But there is nothing unusual about someone who is utr 4.42 beating somone that is in the 5.?? Utr. If they all had over 12 matches (of that type singles/doubles) in the past 12 months and they were blow outs that would be something. But otherwise if the opponents were lower end 5.xx utr it is like a 3.82 ntrp beat a 3.95 ntrp. And if they were 4.0 women then they were likely very low end 5.xx utr. Because 5.00 utr is extreme upper end of womens 4.0 tennis.

 

[TennisOTM]

WTN is weak on the self-rating with over-use of pre-assumptions (like the age penalty), but gains momentum and is probably the most accurate of the rating systems for players that have a substantial amount of data in the system.

Would love to see evidence for this if you have any. When I tested their match predictions last year, WTN actually did worse for players who had more data in the system from the prior year. Maybe it's better now?

TR is pretty solid for rec level play in both singles and doubles, with bonus for separating out mixed. Algo is simple and logical and easy to reverse engineer. It doesn’t separate singles and doubles rating, so can be misleading if you don’t review the match history to estimate those ratings separately yourself. It’s also a bit less sticky than the other rating systems as it only averages last several match ratings, so it has more recency bias and more sensitive to outlier results.

I don't think TR has that much recency bias. TR averages a new match rating with the previous three dynamic ratings, which is stickier than you think. E.g. if you start out with 3 straight dynamic ratings of 3.00, then play 10 matches in a row rated 3.50, your TR dynamic rating will still be under 3.39.

 

[ChaelAZ]

So having not played an official match since 2023, UTR just dropped and set my rating to UTR 5. I guess that makes sense overall. Almost enough for me to do a UTR flex or such just to get it back up....almost.

 

[travlerajm]

So having not played an official match since 2023, UTR just dropped and set my rating to UTR 5. I guess that makes sense overall. Almost enough for me to do a UTR flex or such just to get it back up....almost.

UTR punishes you for being inactive and older than dirt. For juniors, UTR rewards inactivity. It’s a zero sum system that must balance out.

 

[ChaelAZ]

UTR punishes you for being inactive and older than dirt. For juniors, UTR rewards inactivity. It’s a zero sum system that must balance out.

I am active though!

 

[Purestriker]

UTR punishes you for being inactive and older than dirt. For juniors, UTR rewards inactivity. It’s a zero sum system that must balance out.

This is true, I played 3 matches in a week and my rating soared. Now that I have not had an official match in over two weeks it is starting to drop again.

 

[Roforot]

UTR is lame; I'm a NTRP 3.5 player who is admittedly much better at doubles and will likely move up to 4.0, but somehow it has jacked my singles up to 6.5. When I play the 3.5 singles flex league, all my opponents are in the 3-4 range... so basically UTR isn't counting those matches or correcting my jacked up singles rating even though I'm basically playing who I should?? In the end, my singles rating doesn't matter, I'm not going to play much, but it irritates me that it's higher than my doubles UTR...

 

[puppybutts]

I have barely played any UTR tennis this year and my UTR still went up by almost a full 1.0. I thought that was bizarre, but I did also think my UTR was always weirdly low compared to NTRP. i don't really put much worry into my ratings though as long as my opponents remain around the same level.

 

[Roforot]

I have barely played any UTR tennis this year and my UTR still went up by almost a full 1.0. I thought that was bizarre, but I did also think my UTR was always weirdly low compared to NTRP. i don't really put much worry into my ratings though as long as my opponents remain around the same level.

were you playing league matches? then those should still impact your UTR.

 

[puppybutts]

were you playing league matches? then those should still impact your UTR.

sorry to clarify I wasn't playing much tennis at all