Dear UTR: you might want to take a closer look at your algo

[jmnk]

@Moon Shooter - you may want to contact FIDE and tell them that this Elo ranking they use is, well, not great. In the just finished FIDE World Championship Match between Magnus Carlsen and Ian Nepomniachtchi the Elo difference was 74 points. According to what they say a match between players with 74points difference should result in higher ranked player getting about 60% of possible points. Well, Carlsen got a whooping 68% of possible score. And no, 8% difference is _not small_. Per Elo, in order for a player to get 68% of possible points he should have been 135 Elo points higher than the opponent.

Clearly something is not right.



Or, you know, because they actually _play the game_, in any particular match, the outcome may not reflect the ranking formula expectation. Just a thought..........

 

[Moon Shooter]

Jmnk
Are you sure they weren’t 74 points different the day before the match but then at 12:01 am the first day of the match their ratings changed to be 135 points different?

 

[S&V-not_dead_yet]

@Moon Shooter - you may want to contact FIDE and tell them that this Elo ranking they use is, well, not great. In the just finished FIDE World Championship Match between Magnus Carlsen and Ian Nepomniachtchi the Elo difference was 74 points. According to what they say a match between players with 74points difference should result in higher ranked player getting about 60% of possible points. Well, Carlsen got a whooping 68% of possible score. And no, 8% difference is _not small_. Per Elo, in order for a player to get 68% of possible points he should have been 135 Elo points higher than the opponent.

Clearly something is not right.



Or, you know, because they actually _play the game_, in any particular match, the outcome may not reflect the ranking formula expectation. Just a thought..........

I feel for Nepo: getting outplayed in that marathon game 6 was a Carlsen trademark but then he blundered the pawn in game 8 and then the piece in game 9. I'm not sure many predicted this lopsided of an outcome.

 

[jmnk]

I feel for Nepo: getting outplayed in that marathon game 6 was a Carlsen trademark but then he blundered the pawn in game 8 and then the piece in game 9. I'm not sure many predicted this lopsided of an outcome.

yes, exactly. Ian is way better player than he showed during that match. And indeed neither the expert nor the ranking-based logic predicted such a score. And yet it happened. You know, like when a 3.5 plays 4.0 and it is a close match. Also _may_ happen. Doesn't mean the rankings are wrong.

 

[Moon Shooter]

I didn’t watch that match. But I am not surprised by that result. At that level once you are down in a match you can no longer play safe. You have to push to try to create complications and that means taking risks and you may drop even more games. The added stress can also lead to mistakes.

one of the nice things about chess ratings is you can objectively measure an amazing performance. I was a spectator at this tournament in 2014:

Throwback Thursday: Caruana scores seven wins in a row

In a mind-boggling performance, Fabiano Caruana won the second edition of the Sinquefield Cup after scoring victories in his first seven games and obtaining a stratospheric Tournament Performance Rating. The six-player field had an average Elo rating of 2802 and included world champion Magnus...

en.chessbase.com

3103 performance rating!

edit:
And jmnk a 3.5 and 4.0 can have pretty much identical ratings. (With appeals the 3.5 may even be higher) So of course the 3.5 can win. You are arguing with straw men. I never claimed that a remarkable result is proof the ratings are wrong.

If USTA ran the ratings for chess Carlsen and Nepo would be rated identically, Carlsen would win every single game and then someone rom USTA would be on saying that would be a typical result for two people rated exactly the same.

 

[S&V-not_dead_yet]

I didn’t watch that match. But I am not surprised by that result. At that level once you are down in a match you can no longer play safe. You have to push to try to create complications and that means taking risks and you may drop even more games. The added stress can also lead to mistakes.

Everything you write is true but you have the benefit of hindsight to come up with an ex-post facto explanation.

A better test would have been if someone told you the result before the match started.

one of the nice things about chess ratings is you can objectively measure an amazing performance.

How do you define "objectively"? If it's based on the existence of an algorithm, then UTR can be said to be objective also.

That's obviously not enough: you need to know how the algorithm was constructed. No algo is perfect: the designers had to make compromises on what to drop and also how much to weight what they included.

edit:
And jmnk a 3.5 and 4.0 can have pretty much identical ratings. (With appeals the 3.5 may even be higher) So of course the 3.5 can win. You are arguing with straw men. I never claimed that a remarkable result is proof the ratings are wrong.

That wasn't @jmnk's argument, as I interpreted it. You're arguing the 3.5 could win because he is a lot closer in rating to the 4.0. jmnk didn't make that assumption. Maybe the 3.5 had a great day and the 4.0 had a lousy one.

If USTA ran the ratings for chess Carlsen and Nepo would be rated identically, Carlsen would win every single game and then someone rom USTA would be on saying that would be a typical result for two people rated exactly the same.

Their published NTRP would be the same [ie 7.0] but their internal #s would likely show a differential in favor of Carlsen.

And I doubt they would say the result was typical, only that it wasn't unreasonable.

 

[OnTheLine]

It says so right there on the page. I don’t recall it being available before, but maybe it’s new.

Edit: I tested it out, just becsuse. I tried the auto-appeal both up and down, and was denied in both directions.

I’d guess that the auto-appeal allowance threshold is within 0.05 of the edge, or something like that.

Here is what USTA says about this ...

Please review the following points before proceeding to the steps below to appeal your NTRP year-end rating or self-rating:

• If you do not see the option to Appeal Rating Level, this means you will need to complete a NTRP Leagues Self Rate

Has nothing to do with whether you are within range of an appeal or not ...

 

[Moon Shooter]

Everything you write is true but you have the benefit of hindsight to come up with an ex-post facto explanation.

A better test would have been if someone told you the result before the match started.

I did not try to predict this match and have not been following chess lately. It is not going to be the strongest challengers playing for the title as long as they just have a championship tournament to get in. But what I said about having to take more risks if you are down is a pretty standard view.

How do you define "objectively"? If it's based on the existence of an algorithm, then UTR can be said to be objective also.

Both FIDE and UTR are equally objective. The issue is transparency. What was performance rating did Medvedev achieve at the US open? Who knows? (And really when the UTR ratings are capped who cares?) I'm just saying it is nice when it is transparent. Even amateurs can hope to have an event where everything just seems to work out and get a high performance rating. Quantifying that through a performance rating is nice. But you can't do that with NTRP or UTR.

That's obviously not enough: you need to know how the algorithm was constructed. No algo is perfect: the designers had to make compromises on what to drop and also how much to weight what they included.

Thats right it should be transparent to do that. Chess ratings are transparent. Tennis associations act like they have some super secret formula that they need to protect when really they are just using the same sort of systems that have been around for decades. It's really silly.

That wasn't @jmnk's argument, as I interpreted it. You're arguing the 3.5 could win because he is a lot closer in rating to the 4.0. jmnk didn't make that assumption. Maybe the 3.5 had a great day and the 4.0 had a lousy one.

JMNK was arguing with a strawman that thought any remarkable result would be proof the rating system is broken. My point was just that USTA's system is practically meaningless. If I say I am a 3.5 rated player that tells you pretty much nothing about my ability compared to about half of the adult rec tennis players.

 

[S&V-not_dead_yet]

Thats right it should be transparent to do that. Chess ratings are transparent. Tennis associations act like they have some super secret formula that they need to protect when really they are just using the same sort of systems that have been around for decades. It's really silly.

It's not transparent for the same reason the recipe for Coke is not transparent: there is $ involved.

JMNK was arguing with a strawman that thought any remarkable result would be proof the rating system is broken. My point was just that USTA's system is practically meaningless. If I say I am a 3.5 rated player that tells you pretty much nothing about my ability compared to about half of the adult rec tennis players.

I don't think it's practically meaningless at all. I'm OK with the lack of transparency. Why? Because I get competitive matches, for the most part.

Yes, I live in a dense, tennis hotspot so that skews my outlook. If you're not getting competitive matches because you're rated too low or too high, at least you can appeal. The worst-case would be if you win too easily and lose too easily the majority of the time, which implies the cohort is too large [I believe you've discussed that before and I have no remedy for that].

 

[Moon Shooter]

It's not transparent for the same reason the recipe for Coke is not transparent: there is $ involved.

By not making it transparent in the way Chess does USTA and UTR are losing money. The transparency adds to legitimacy. That is why so many people post here in the forums about how NTRP and UTR seem inaccurate because this person got a bump or this rating is such and such. This leads people to not care about the rating system. If people don't care about the rating service then that service is *less* valuable not more valuable.

Arpad Elo, and Mark Glickman already developed the system. The different rating systems (in chess and tennis) are just making minor tweaks that likely make it more or less accurate. Making it seem like they have some patent worthy new invention just adds a layer of dishonesty, to the stupidity of hiding this information.

I don't think it's practically meaningless at all. I'm OK with the lack of transparency. Why? Because I get competitive matches, for the most part.

Its not that hard to get "compatible" matches even without their rating system. In many cases the rating system actually makes for less competitive matches because a 3.49 is put in a different league than a 3.52. Someone just looking at the players would probably see they are more compatible than putting the 3.49 with a 3.03 or putting the 3.52 with a 3.97.

Yes, I live in a dense, tennis hotspot so that skews my outlook. If you're not getting competitive matches because you're rated too low or too high, at least you can appeal. The worst-case would be if you win too easily and lose too easily the majority of the time, which implies the cohort is too large [I believe you've discussed that before and I have no remedy for that].

No the worst case is not enough people in the area finds value joining the official organization for US tennis and so I get few if any matches through that organization.


There are plenty of things USTA or UTR could do you just keep denying all the problems because they seem to work for you in your area. Your constant response to just about every issue seems to be everything is fine and anyone complaining is wrong "Because I get competitive matches, for the most part." Your failure to even try to consider how other people are effected by these rules is pretty astounding.

Edit: BTW I am not saying UTR or USTA claims they are doing anything deserving of a patent or intellectual property right protection. I was just pointing out that the intellectual work of creating ratings has been out for a long time now.

 

[travlerajm]

By not making it transparent in the way Chess does USTA and UTR are losing money. The transparency adds to legitimacy. That is why so many people post here in the forums about how NTRP and UTR seem inaccurate because this person got a bump or this rating is such and such. This leads people to not care about the rating system. If people don't care about the rating service then that service is *less* valuable not more valuable.

Arpad Elo, and Mark Glickman already developed the system. The different rating systems (in chess and tennis) are just making minor tweaks that likely make it more or less accurate. Making it seem like they have some patent worthy new invention just adds a layer of dishonesty, to the stupidity of hiding this information.




Its not that hard to get "compatible" matches even without their rating system. In many cases the rating system actually makes for less competitive matches because a 3.49 is put in a different league than a 3.52. Someone just looking at the players would probably see they are more compatible than putting the 3.49 with a 3.03 or putting the 3.52 with a 3.97.



No the worst case is not enough people in the area finds value joining the official organization for US tennis and so I get few if any matches through that organization.


There are plenty of things USTA or UTR could do you just keep denying all the problems because they seem to work for you in your area. Your constant response to just about every issue seems to be everything is fine and anyone complaining is wrong "Because I get competitive matches, for the most part." Your failure to even try to consider how other people are effected by these rules is pretty astounding.

Edit: BTW I am not saying UTR or USTA claims they are doing anything deserving of a patent or intellectual property right protection. I was just pointing out that the intellectual work of creating ratings has been out for a long time now.

Agree 100%. There really isn’t any value in concealing the algorithm. And people would pay more attention to their UTR if the rating system were transparent, because then it would be viewed as more trustworthy.

 

[S&V-not_dead_yet]

By not making it transparent in the way Chess does USTA and UTR are losing money.

Likely they have considered your points and many others and decide on the status quo. If you want to make your case to them that they are losing money, you're free to do so.

The transparency adds to legitimacy. That is why so many people post here in the forums about how NTRP and UTR seem inaccurate because this person got a bump or this rating is such and such. This leads people to not care about the rating system. If people don't care about the rating service then that service is *less* valuable not more valuable.

No matter the system, you will come up with edge cases. The question isn't whether the edge cases exist but are they a significant % of the total?

[quote
Its not that hard to get "compatible" matches even without their rating system. In many cases the rating system actually makes for less competitive matches because a 3.49 is put in a different league than a 3.52. Someone just looking at the players would probably see they are more compatible than putting the 3.49 with a 3.03 or putting the 3.52 with a 3.97.
[/quote]

You have to draw the line somewhere. How big of a rating band do you need to include the 3.52? Then what about the 3.53? And 3.55? And 3.60?

Instead of just looking at one match, look at all of the matches the arbitrary NTRP and the results then compare them to what the results would have been with a larger band.

And if you make your bands smaller, are you going to get critical mass? Having a tournament with only 2 people in a class might be considered a success from a matchup standpoint because the outcome was close but it would be a failure from a tournament draw perspective.

No the worst case is not enough people in the area finds value joining the official organization for US tennis and so I get few if any matches through that organization.

There are plenty of things USTA or UTR could do you just keep denying all the problems because they seem to work for you in your area. Your constant response to just about every issue seems to be everything is fine and anyone complaining is wrong "Because I get competitive matches, for the most part." Your failure to even try to consider how other people are effected by these rules is pretty astounding.
[/QUOTE]

I admit that I'm not taking others into account. Again, it works for me. And I suspect it works for a significant majority. If it doesn't work for you, then you have to figure out what to do. Since complaining about it isn't causing anything to change, the only thing you can do in the short-run is to change your behavior. If you're unwilling to do that, then you are at an impasse.

It seems you'd be better off dropping out of the system altogether and just building a network of people who are close in skill range. The benefits of the system seem to be outweighed by the drawbacks.

 

[S&V-not_dead_yet]

Its not that hard to get "compatible" matches even without their rating system. In many cases the rating system actually makes for less competitive matches because a 3.49 is put in a different league than a 3.52. Someone just looking at the players would probably see they are more compatible than putting the 3.49 with a 3.03 or putting the 3.52 with a 3.97.

BTW: the same problem exists for Elo. If I'm rated 1800, I have to enter the A bracket where I could play a 1999, who is also an A player. I will get demolished. Does that mean Elo is worthless? Not at all. It just means I need to improve to be competitive with that rating. And if I happen to drop to 1799, all of a sudden I can enter as a B player where I will crush the 1600.

But those are edge cases. More are to be expected if I have a rating that's close to the edge of any boundary. But the vast majority of players will be a lot closer to the median [95% within +/- 3 sigmas of the median].

 

[jmnk]

BTW: the same problem exists for Elo. If I'm rated 1800, I have to enter the A bracket where I could play a 1999, who is also an A player. I will get demolished. Does that mean Elo is worthless? Not at all. It just means I need to improve to be competitive with that rating. And if I happen to drop to 1799, all of a sudden I can enter as a B player where I will crush the 1600.

But those are edge cases. More are to be expected if I have a rating that's close to the edge of any boundary. But the vast majority of players will be a lot closer to the median [95% within +/- 3 sigmas of the median].

Of course. The fact that those artificial bands exist, and they have to for practical reasons (you call then 3.5/4.0/4.5 ranking in tennis, and A/B/C tournament brackets or category in chess) in pretty much any recreational competition is pretty obvious to anyone except @Moon Shooter apparently.

 

[Moon Shooter]

S&V and JMNK

You both still seem to think the only use of a rating system is to set up leagues. That is how USTA looks at it but they don't get anything close to the full benefit of having a rating system. People play rated competitive chess all over the country not just in chess hotbeds like New York. They play it in tiny towns and all over the country. Why? Because there is a rating system that is an objective measure of how you are doing at chess and competitive people want to know how they compare to others.

A rating system can give additional meaning to every rated game. In every rated game you are not just getting a score against a single opponent you are getting a score against the rest of the country or even the world.

This is why you don't understand how chess ratings work. In chess very few tournaments are divided by class. Yes most of the big money tournaments are done that way but they only make up a small fraction of rated chess games. The vast majority are just local tournament or ladders. The swiss style format will get the better players playing against each other - you don't need a rating system. This is why your example is all but irrelevant to activities like chess that properly use a rating system.

But ratings and increasing your rating is a an additional reason to play in these tournaments. If you do well you gain rating points and you have an objective achievement of strength at that activity. Because the USTA rating system is so hidden and vague it can't get people interested in tournaments. Of course they can't get interest in tournaments with this terrible vague and opaque rating system! What does winning a tournament mean? Well it depends how good the people were right? Well if you have a legitimate and transparent rating system you know. If you don't have that, then who knows what it means to win?

But this is the important thing - people play competitively not only or even primarilly to win tournaments - of course someone who is 1300 will know they are not going to win a tournament with 4 people rated over 1600, The achievement that these tournaments provide is to gain rating points. So even if they lose four out of their six games to stronger opponents they still get experience have fun and may gain some rating points if they do better than expected.

USTA has this horribly watered down rating system and then they wonder why tournaments are not doing so well. Well what sort of tennis achievement would anyone get if they pay money to play in a USTA tournament? If they are only looking at the possibility of winning the tournament you probably won't get enough players to even have the tournament. If adult rec tennis doesn't have a good rating system it will never have a decent tournament culture.

 

[travlerajm]

Update:
My UTR climbed 0.5 UTR points this morning.
Feels like it’s my birthday.

 

[J011yroger]

Update:
My UTR climbed 0.5 UTR points this morning.
Feels like it’s my birthday.

At this rate you will be on tour by Christmas!

J

 

[travlerajm]

At this rate you will be on tour by Christmas!

J

I like my chances.

 

[Moon Shooter]

Likely they have considered your points and many others and decide on the status quo. If you want to make your case to them that they are losing money, you're free to do so.



No matter the system, you will come up with edge cases. The question isn't whether the edge cases exist but are they a significant % of the total?

Its not that hard to get "compatible" matches even without their rating system. In many cases the rating system actually makes for less competitive matches because a 3.49 is put in a different league than a 3.52. Someone just looking at the players would probably see they are more compatible than putting the 3.49 with a 3.03 or putting the 3.52 with a 3.97.

You have to draw the line somewhere. How big of a rating band do you need to include the 3.52? Then what about the 3.53? And 3.55? And 3.60?

Instead of just looking at one match, look at all of the matches the arbitrary NTRP and the results then compare them to what the results would have been with a larger band.

And if you make your bands smaller, are you going to get critical mass? Having a tournament with only 2 people in a class might be considered a success from a matchup standpoint because the outcome was close but it would be a failure from a tournament draw perspective.

No you don't need to draw lines for many potential adult rec tennis activities - including ladders or tournaments. The problem is tournaments will never be a good format as long as USTA guts the relevance of ratings by hiding important information. USTA could leave it up to local tournament directors as to how they want to draw lines but again the same problem exists. The hide the important information. Also USTA could allow some leagues to draw lines differently and not based on gender which would make the leagues much more flexible, allow smaller bands/increased competitiveness and increase the number of leagues smaller communities could put together etc etc. You just plug your ears and pretend these policies put in place by USTA have no effect on the problems. You may not care about the rest of the country but USTA should, because they are supposed to be the tennis organization for the whole country not just the south and Pacific Northwest.

I admit that I'm not taking others into account. Again, it works for me. And I suspect it works for a significant majority. If it doesn't work for you, then you have to figure out what to do. Since complaining about it isn't causing anything to change, the only thing you can do in the short-run is to change your behavior. If you're unwilling to do that, then you are at an impasse.

It seems you'd be better off dropping out of the system altogether and just building a network of people who are close in skill range. The benefits of the system seem to be outweighed by the drawbacks.

At least you admit you are just concerned how it works for you and are not considering the rest of the country.

And you suspect USTA "works for a significant majority" Significant majority of what exactly? Adults that play rec tennis? What percentage of adult rec tennis players even played USTA tennis last year? Only a fraction of tennis players I know played USTA tennis last year. Are you thinking USTA is working fine for all these adults that choose not to play USTA tennis? This is what I mean about whether USTA has any goals or measures they are trying to use to determine if their policies work.

Are you thinking that USTA works for the people that choose to be in USTA? So as soon as people leave their view no longer counts? Of course then as long as there is one league in the country where more than 60% of the participants are happy to join USTA we can claim "it works for a significant majority." And that way we can then ignore the fact that participation was on the decline even before covid. The fact that more families have access to courts and sports like football are on a decline relative to other sports because of fear of injury would make you think USTA participation should be increasing right?


Dropping out of the few matches USTA offers is not really going to be a solution for me not getting enough matches. I enjoy league play, so it is odd that you would even suggest that.

I find it interesting that you continue to think it is appropriate to tell others what to do. I will continue to propose solutions to what I think are problems with USTA adult rec tennis and listen to people that want to discuss whether those proposals would be good or bad. You can keep repeating that the current system works fine for you in your local tennis hotspot. Whether anything leads to any changes or not, is outside of my control, so I don't worry about it.

 

[jmnk]

looks like another thread was essentially killed by a long post arguing that USTA/UTR is bad, and not publishing an exact algorithm makes people not play tennis. Oh well, not the first time it happened, I'm afraid not the last one either........

 

[travlerajm]

Update:
This morning my UTR plummeted 2.5 UTR points. Back to slumming with the 3.5 ladies.

 

[J011yroger]

Update:
This morning my UTR plummeted 2.5 UTR points. Back to slumming with the 3.5 ladies.

Narrator: He would not be on the tour by Christmas.

J

 

[travlerajm]

Narrator: He would not be on the tour by Christmas.

J

I checked my UTR again this morning. Still dropping like flies high on CO2. I’m officially a UTR 4 again.

So, to recap: Since my last match played, my UTR has moved from 4, to 6, to 8, to 7, to 5, and back to 4 again.

Shouldn’t a rating system claiming to be “the world’s most accurate rating system” be able to differentiate skill level between a perennial high 4.5 ntrp and my 60-year-old 3.5 ntrp female mixed partner? Or is that asking too much?

And why does it think I’m a 3.5 ntrp lady some days, but a low 5.0 ntrp other days? Is UTR moody?

 

[S&V-not_dead_yet]

I checked my UTR again this morning. Still dropping like flies high on CO2. I’m officially a UTR 4 again.

So, to recap: Since my last match played, my UTR has moved from 4, to 6, to 8, to 7, to 5, and back to 4 again.

Shouldn’t a rating system claiming to be “the world’s most accurate rating system” be able to differentiate skill level between a perennial high 4.5 ntrp and my 60-year-old 3.5 ntrp female mixed partner? Or is that asking too much?

And why does it think I’m a 3.5 ntrp lady some days, but a low 5.0 ntrp other days? Is UTR moody?

Maybe your DNTRP is varying just as much but because you can't see, you aren't aware. tennisrecord.com gives match ratings as well as overall ratings: mine swings quite a bit, from 4.04 to 4.63. I think most people have quite a large spread between max and min because we are inconsistent.

 

[travlerajm]

Maybe your DNTRP is varying just as much but because you can't see, you aren't aware. tennisrecord.com gives match ratings as well as overall ratings: mine swings quite a bit, from 4.04 to 4.63. I think most people have quite a large spread between max and min because we are inconsistent.

My UTR is yo-yo-ing between UTR 4 and UTR 8 every 3 days during my period of inactivity? How can I blame that on my inconsistency? I’ve been consistently not playing any matches during the last 2 weeks. Obviously there is a serious bug in the algorithm, and I’m puzzled why some folks seem so hellbent on coming to its defense.

 

[S&V-not_dead_yet]

My UTR is yo-yo-ing between UTR 4 and UTR 8 every 3 days during my period of inactivity? How can I blame that on my inconsistency? I’ve been consistently not playing any matches during the last 2 weeks. Obviously there is a serious bug in the algorithm, and I’m puzzled why some folks seem so hellbent on coming to its defense.

My point was that since you can't see your DNTRP, you don't know if it's also bouncing around.

I have no skin in the game. I'm just pointing out other explanations or possibilities. Maybe you've hit some sort of "dead zone" in the algorithm. Yes, it decreases one's confidence in the accuracy. If I depended on it being accurate, I'd make sure to play enough matches so it wasn't bouncing around as much. My UTR rose 200 points while I was idle; go figure.

 

[jmnk]

My UTR is yo-yo-ing between UTR 4 and UTR 8 every 3 days during my period of inactivity? How can I blame that on my inconsistency? I’ve been consistently not playing any matches during the last 2 weeks. Obviously there is a serious bug in the algorithm, and I’m puzzled why some folks seem so hellbent on coming to its defense.

Again, it says right there " Player metrics are not reliable because player does not have enough match results. ". What difference does it make then if that still unreliable ranking says you are 7.55 or 4.74 or 10.25? Obviously there's a serious lack of data feeding the algorithm, and I'm puzzled why some folks seem so hellbent on checking such unreliable ranking that often instead of, like, playing matches.

 

[Moon Shooter]

Again, it says right there " Player metrics are not reliable because player does not have enough match results. ". What difference does it make then if that still unreliable ranking says you are 7.55 or 4.74 or 10.25? Obviously there's a serious lack of data feeding the algorithm, and I'm puzzled why some folks seem so hellbent on checking such unreliable ranking that often instead of, like, playing matches.

JMNK UTR sucks because it excludes too much data.

It is not that people are hellbent on checking their rating "instead of, like, playing matches." It is that they are recognizing UTR is unlikely to be a good rating system for adult rec tennis and it would be nice if UTR adjusted their algo so it would be.

 

[jmnk]

JMNK UTR sucks because it excludes too much data.

It is not that people are hellbent on checking their rating "instead of, like, playing matches." It is that they are recognizing UTR is unlikely to be a good rating system for adult rec tennis and it would be nice if UTR adjusted their algo so it would be.

Which matches did UTR exclude in OP's case? It included mixed doubles which, and I hope even you can agree, is the least reliable kind of a tennis match there is because the players involved vary in ranking so much. It included matches where some of the players involved do not have established reliable ranking. If anything it included _too many_ matches in OP's case - but, well, these were almost the only matches there were.

Why should UTR adjust its algorithm since it works for everybody, adult rec player including, that plays a reasonable number of matches per year? If you/OP do not like the algorithm - don't look at the ranking it gives.

 

[GrassCourtFan]

Why should UTR adjust its algorithm since it works for everybody, adult rec player including, that plays a reasonable number of matches per year?

Because lots of adult rec players don't play the number of matches that UTR considers "reasonable". Which makes their ratings unreliable and erratic.

Which is fine if UTR doesn't really need to work for adult rec players! If it's a rating system primarily for mid to high level juniors, there's no need for it to work for players of other demographics. ...but they do call it "universal". Which doesn't seem accurate.

 

[jmnk]

Because lots of adult rec players don't play the number of matches that UTR considers "reasonable". Which makes their ratings unreliable and erratic.

Could you please describe in 5 sentences or less an idea for a ranking system that gives reliable results based on less than 5 matches? Or any example of such system existing in practice? (UTR says that they need about 5-10 matches to give reliable ranking. If you can't manage to play that extraordinary number of 5 matches per year I would wonder first why you need to know your ranking in the first place......)

Which is fine if UTR doesn't really need to work for adult rec players! If it's a rating system primarily for mid to high level juniors, there's no need for it to work for players of other demographics. ...but they do call it "universal". Which doesn't seem accurate.

yes, UTR is wrong because they did not put a disclaimer that their system is universal _only if a player actually plays matches_. Yap, that is not that easy to figure out, and should be indeed disclosed. I would go even further - anyone that paid for a premium account should sue them because they did not deliver on the promise of 'universal'. Like my kids are not even ranked there, not even with 'unreliable' ranking. I mean they never played a single tennis match but UTR says its universal so I think I have a case.... I mean are you serious??

 

[GrassCourtFan]

Could you please describe in 5 sentences or less an idea for a ranking system that gives reliable results based on less than 5 matches?

Any system that keeps matches for more than a year, thus allowing players who play 3-5 rated matches per year to get accurate ratings because they have more matches over several years.

Or any example of such system existing in practice?

USTA NTRP, which rated me perfectly accurately after a year where I’d played zero rated matches (2020), since it just kept my rating from the year before.

(UTR says that they need about 5-10 matches to give reliable ranking. If you can't manage to play that extraordinary number of 5 matches per year I would wonder first why you need to know your ranking in the first place......)

Same reasons someone who plays a lot of rated matches wants to know their ratings.

(Also, you’re missing one big distinction - rated matches vs all matches. Lots of people who play only a few rated matches a year play plenty more matches the rating system doesn’t know about.)

yes, UTR is wrong because they did not put a disclaimer that their system is universal _only if a player actually plays matches_.

Nah, lots of people who play a lot of tennis are mis-rated by UTR.

Of the people on the USTA teams I played on last year, I think a majority had 5 or fewer rated matches per year. If a rating can’t handle that amount of play, it does not work for adult rec players.

Are you claiming that it’s impossible to rate adult rec players because they mostly don’t play enough? That’s nonsense.

 

[S&V-not_dead_yet]

Are you claiming that it’s impossible to rate adult rec players because they mostly don’t play enough? That’s nonsense.

I'd argue it's not black and white. More like "the fewer matches played, the potentially less accurate the rating and the larger the margin of error."

 

[S&V-not_dead_yet]

My singles UTR just jumped some 200 points again [2nd time it's happened since the pandemic began] with no matches played in over a year. While I was playing singles regularly, it was in the low 7s.

So yeah, weird things happen with low or no data points. Since I can't see the algorithm [and even if I did would I understand it?], my conclusion is that it's for entertainment value. Good thing I don't have to use it to enter a tournament because I'd get laughed [and hit] off of the court.

Pre-pandemic, it worked just fine [no unexplainable huge quantum jumps].

 

[travlerajm]

My UTR just jumped 3 UTR points since yesterday. Back on track for my pro career.

 

[FuzzyYellowBalls]

My UTR just jumped 3 UTR points since yesterday. Back on track for my pro career.

All right, I have to bite, no way that is possible? Has anyone else experienced this ever? I mean mine dipped in the great male singles UTR shift earlier this year, almost 1 point, but you can't even get a rating to move more than .2 or .3 even if you win 2 matches 6-0 6-0. What is happening? Do you have 1 match or 2 or are you looking at singles ratings and doubles? You should email UTR, they would probably like to know if you rating is really jumping around like that.

 

[travlerajm]

All right, I have to bite, no way that is possible? Has anyone else experienced this ever? I mean mine dipped in the great male singles UTR shift earlier this year, almost 1 point, but you can't even get a rating to move more than .2 or .3 even if you win 2 matches 6-0 6-0. What is happening? Do you have 1 match or 2 or are you looking at singles ratings and doubles? You should email UTR, they would probably like to know if you rating is really jumping around like that.

My UTR was lower than my NTRP rating yesterday. So the 3 point jump didn’t exactly put my my UTR into the ATP pro range.

I played 2 mixed matches, the last one a couple of weeks ago. Since the last time I played, my UTR has oscillated up and down on an almost daily basis with a 4 UTR point amplitude (as low as 3.0 NTRP and as high as 5.0 NTRP equivalent, the far extremes of the rec player level spectrum).

Meanwhile, USTA, TennisRecord, TLS, and @schmke seem to be in consensus that I was a high 4.5 yesterday, a high 4.5 today, and probably still a high 4.5 tomorrow.

 

[FuzzyYellowBalls]

My UTR was lower than my NTRP rating. So the 3 point jump didn’t exactly put my my UTR into the ATP pro range.

I played 2 mixed matches, the last one a couple of weeks ago. Since the last time I played, my UTR has oscillated up and down on an almost daily basis with a 4 UTR point amplitude (as low as 3.0 NTRP and as high as 5.0 NTRP equivalent, the far extremes of the rec player level spectrum).

Meanwhile, USTA, TennisRecord, TLS, and @schmke seem to be in consensus that I was a high 4.5 yesterday, a high 4.5 today, and probably still a high 4.5 tomorrow.

Im trying to draw attention to the undeniable fact the UTR algorithm is F’d up. Dear UTR: fix it and I’ll stop.

I need help understanding. So when I read that your UTR jumped 3 UTR points, you are saying your doubles only, because doubles and singles ratings are usually different, your doubles UTR rating went from (I'm just making up numbers to understand) an 8 to a 5 in one day? Then a 4 UTR point amplitude means it went from a 4 back to an 8 UTR? That's why I am confused because that would be some kind of UTR glitch that only you are experiencing, no one else has ever had their UTR go up or down 3 full UTR points or 4 full UTR points in one day.

Did you have only one match on record then lose to lower UTR 0-6 0-6 twice (exaggerating) to give the new matches a huge weight in the calculation?

 

[travlerajm]

I need help understanding. So when I read that your UTR jumped 3 UTR points, you are saying your doubles only, because doubles and singles ratings are usually different, your doubles UTR rating went from (I'm just making up numbers to understand) an 8 to a 5 in one day? Then a 4 UTR point amplitude means it went from a 4 back to an 8 UTR? That's why I am confused because that would be some kind of UTR glitch that only you are experiencing, no one else has ever had their UTR go up or down 3 full UTR points or 4 full UTR points in one day.

Did you have only one match on record then lose to lower UTR 0-6 0-6 twice (exaggerating) to give the new matches a huge weight in the calculation?

I haven’t played any USTA singles in a few years, so I only have a doubles rating. Yes, my doubles UTR has been yo-yo-ing up and down from as high as UTR 8 to as low as UTR 4 multiple times within the past two weeks, despite the fact that I haven’t played during that span.

I’m quite sure it’s not just me. A friend texted me today to tell me that his friend who was the #1 ranked 18yo in the East Section who went undefeated in 5.0 league singles is sporting a UTR 5 (same as average 3.5 ntrp level).

 

[FuzzyYellowBalls]

I haven’t played any USTA singles in a few years, so I only have a doubles rating. Yes, my doubles UTR has been yo-yo-ing up and down from as high as UTR 8 to as low as UTR 4 multiple times within the past two weeks, despite the fact that I haven’t played during that span.

I’m quite sure it’s not just me. A friend texted me today to tell me that his friend who is the #1 ranked 18yo in the East Section who went undefeated in 5.0 league singles is sporting a UTR 5 (same as average 3.5 ntrp level).

Well, I am pretty sure it's just you who is swinging from 8 to 4 in a day, for sure. Having a 5.0 rated as a UTR 5 isn't unheard of, now you wouldn't be alone if that 5.0 rated 5 was and 8 or 9 one day and the next day a 5 then back again, but if he was just a 5.0 at a 5 it can happen, especially with whatever they did to cause the shift this year during covid, some kind of male adult UTR reset or something. If you care to, you can email UTR and ask them, I'd be curious what they would say. Something strange is going on there lol.

 

[S&V-not_dead_yet]

Well, I am pretty sure it's just you who is swinging from 8 to 4 in a day, for sure. Having a 5.0 rated as a UTR 5 isn't unheard of, now you wouldn't be alone if that 5.0 rated 5 was and 8 or 9 one day and the next day a 5 then back again, but if he was just a 5.0 at a 5 it can happen, especially with whatever they did to cause the shift this year during covid, some kind of male adult UTR reset or something. If you care to, you can email UTR and ask them, I'd be curious what they would say. Something strange is going on there lol.

I just looked and my singles UTR jumped 2 points without me playing any singles matches...and this giant jump has happened once before [both during the pandemic when I've played no rated singles matches]. I'm not even sure how I could have a singles UTR since I don't have any rated matches for over a year but there you have it.

The last time I looked a few weeks ago, I didn't have a singles UTR, I thought because of lack of matches. But that apparently hasn't stopped the computer from granting me one.

I'm pretty sure if I emailed them, the reply would be "play more matches".

 

[travlerajm]

Until the algorithm problems are fixed, I am going to boycott any further discussion of UTR as a rating system for rec players. It is pretty much useless when we we already have TR and TLS working just fine.

For pros and juniors, it still seems to have some utility.

 

[FuzzyYellowBalls]

I just looked and my singles UTR jumped 2 points without me playing any singles matches...and this giant jump has happened once before [both during the pandemic when I've played no rated singles matches]. I'm not even sure how I could have a singles UTR since I don't have any rated matches for over a year but there you have it.

The last time I looked a few weeks ago, I didn't have a singles UTR, I thought because of lack of matches. But that apparently hasn't stopped the computer from granting me one.

I'm pretty sure if I emailed them, the reply would be "play more matches".

Only one way to find out. Do you mean you have never played a singles match and it gave you a rating or it moved while you didn't play a match, but you played one a year ago that it picked up? And your rating went from say 5.12 to 7.12 or it went from 5.4 to 7.1, and that is "2" points?

 

[FuzzyYellowBalls]

Until the algorithm problems are fixed, I am going to boycott any further discussion of UTR as a rating system for rec players. It is pretty much useless when we we already have TR and TLS working just fine.

For pros and juniors, it still seems to have some utility.

It's the best rating system, UTR, for calculating lineups and strategy for sectionals and nationals because it captures all the matches and ratings of opponents for all the sandbagging, like combo, non usta tournaments, open tournaments that don't count for USTA ratings, little overseas matches for those who move here and sandbag, all that kind of stuff. So, it has a huge advantage in that area. Plus it's good when you live in an area where there are 3-7 UTR tournaments a month that adults participate in.

 

[S&V-not_dead_yet]

Only one way to find out. Do you mean you have never played a singles match and it gave you a rating or it moved while you didn't play a match, but you played one a year ago that it picked up? And your rating went from say 5.12 to 7.12 or it went from 5.4 to 7.1, and that is "2" points?

I've played plenty of rated singles matches but none in the past year.

It went up about 2.00 points [5.12 to 7.12] but that's based on what I remember my singles UTR being because recently, the page didn't display one [presumably because of insufficient data].

 

[jmnk]

Until the algorithm problems are fixed, I am going to boycott any further discussion of UTR as a rating system for rec players.

Let me fix that for you: Until the algorithm problems are fixed statisticians find a way to reliably predict results based on a vast sample of less than 5 matches, I am going to boycott any further discussion of UTR as a rating system for rec players players that play no more than 5 matches a year.

It is pretty much useless when we we already have TR and TLS working just fine.

Meanwhile, USTA, TennisRecord, TLS, and @schmke seem to be in consensus that I was a high 4.5 yesterday, a high 4.5 today, and probably still a high 4.5 tomorrow.

That's interesting take given the following facts:

• There's no way to tell from USTA whether you are high, medium, or low within a given level.
• Your TLS ranking appears to be 0.2 lower than TR one, so it is neither consistent with TR, nor does it seem to suggest you are 'high' 4.5.
• Your TR ranking is higher than that of our beloved @GSG - who did beat you fairly convincingly.
• As a side note - neither TR nor TLS has our beloved Mark Sansait in 5.0 and above range - while he is 5.0 per USTA. If one thinks that Mark is not 5.0 player then he needs to have his opinion adjusted.

The point is that neither TR nor TLS are 'working just fine' - they can't do wonders without play data either.

 

[travlerajm]

Update:
Pleased to report that my UTR has jumped for a third consecutive day and is nearly up to the UTR 8 starting point for the year. Time to get my pro spec racquets ready for 2022.

 

[andfor]

I'll bite. My UTR updated again today. I've pingponged from 6.85 to 7.39, mostly staying in the low 7's since I started checking it earlier this year. My rating is doubles only (16 matches). I've go no complaints as my low 7 rating seems consistent with my low 4.5 rating.

Despite knowing that most of us rec players ratings have a degree of inaccuracy for those who have little match data reported, I find the UTR and TR systems excellent for arranging team tennis line ups.

I was going to post much longer, but decided to call for our friends at Universal Tennis Rating to jump in. In the past they have weighed in with some helpful posts. Hopefully they will contribute to the conversation here.

 

[travlerajm]

Update:
My UTR climbed another 6 UTR points today:

My UTR is going to Explode in 2022!!!!

With my UTR stalling out this year in a fluctuating mess, and all indications telling me that the rating system works much better for juniors and pros than it does for adult rec players, I’ve decided to pursue a novel approach to boost my UTR: Yes! I’ve rounded up a motley band of talented...

tt.tennis-warehouse.com

 

[FuzzyYellowBalls]

Sooo, your UTR is now 10 or 14? I'm so confused