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

[travlerajm]

I just played my first USTA match in nearly 2 years.
8.0 mixed.
Yeah. I know I played like crap. I was rusty from not touching racquet in a week. I choked a few points in the super.

But…

After the match, I looked on UTR site, and my UTR has been updated to account for the new data.

My UTR has been downgraded… A lot. We played against a team with combined 11 UTR. Won equal games. My partner was a UTR 4. The other 3 players on court having been playing frequently this year so lots of data on them. Shouldn’t my match rating be a 7 then ?

UTR now says I’m a 4. A UTR 4.xx with 2 decimal place accuracy. Same as my female 3.5 mixed partners. Is it the rating decay with inactivity thing?

TennisRecord says I’m still about 4.4 NTRP (high 4.5) same as I’ve been for the last few years. TLS says I’m a 4.4 too based on my early 2021 results. And the USTA says my computer rating is in range of appealing up to 5.0. And the last year I played a few usta matches, in 2019, I made the TT 4.5 all-star team, which meant schmke had me at least 4.41.

But of course, UTR is the world’s most accurate algorithm. So maybe they know something others don’t?

I mean… I’m ok being a 4. This way my opponents won’t expect much. But UTR, just saying, you might want to double-check your algo and make sure your coder didn’t leave a glitch in there.

 

[schmke]

Well, I'm not playing you now! Getting beat by a 4 would tank my UTR!

 

[BallBag]

It's a scheme to get you to play more so you won't have that blight on your name.

 

[atatu]

Just out of curiosity what was your UTR before this match ?

 

[travlerajm]

Just out of curiosity what was your UTR before this match ?

If I recall correctly, they had me most recently as an 8 for doubs, UR for singles.

I had previously posted that my singles UTR drifted from 9 to 7 to 5 to UR during the course of the year in 2019 during my period of inactivity without me playing a match. What’s going on here? All I ask for is a rating system that makes sense.

 

[S&V-not_dead_yet]

If I recall correctly, they had me most recently as an 8 for doubs, UR for singles.

I had previously posted that my singles UTR drifted from 9 to 7 to 5 to UR during the course of the year in 2019 during my period of inactivity without me playing a match. What’s going on here? All I ask for is a rating system that makes sense.

It obviously deals less well with a small # of data points. It's tough to invoke the Law of Large Numbers with only one sample.

There's also the question of recency: I assume more recent matches have a greater weight than more distant ones.

As a high 4.5, the algo might have expected you to not only win an 8.0 match against a combo 11 UTR but possibly even dominate.

I could see your *match* UTR being 4 but for your overall UTR to drop from 8 to 4 seems like a huge move. Call it the Law of Small Numbers.

I wouldn't lose much sleep over what effect one match had unless it ruined my plans to compete at a certain UTR in the near-future.

 

[travlerajm]

It obviously deals less well with a small # of data points. It's tough to invoke the Law of Large Numbers with only one sample.

There's also the question of recency: I assume more recent matches have a greater weight than more distant ones.

As a high 4.5, the algo might have expected you to not only win an 8.0 match against a combo 11 UTR but possibly even dominate.

I could see your *match* UTR being 4 but for your overall UTR to drop from 8 to 4 seems like a huge move. Call it the Law of Small Numbers.

I wouldn't lose much sleep over what effect one match had unless it ruined my plans to compete at a certain UTR in the near-future.

Of course, after my rating recalibration, my UTR is now listed as a 4 for all of my matches in the system. So even when I played matches where TR spit out 5.0 ratings, UTR still has me as a 4 (equivalent to a 3.0 ntrp).

This doesn’t seem to have anything to do with small data set. It seems like it exposes that the assumption that rating should decay with inactivity is very flawed.

I suspect that it assumes that a high 4.5 becomes a 3.0 if he sits out a year of usta.

 

[BallBag]

Of course, after my rating recalibration, my UTR is now listed as a 4 for all of my matches in the system. So even when I played matches where TR spit out 5.0 ratings, UTR still has me as a 4 (equivalent to a 3.0 ntrp).

On the bright side, your current rating of 4 now shows up on the record of all your previous opponents. So somewhere out there there is UTR 8 with a 6-3 6-4 loss to a UTR 4.

 

[travlerajm]

On the bright side, your current rating of 4 now shows up on the record of all your previous opponents. So somewhere out there there is UTR 8 with a 6-3 6-4 loss to a UTR 4.

I am still coming to terms with my 3.5 mixed partners having the same skill rating as me.

 

[BallBag]

I am still coming to terms with my 3.5 mixed partners having the same skill rating as me.

Yes but imagine how good she must feel having the same skill rating as you.

 

[J011yroger]

Of course, after my rating recalibration, my UTR is now listed as a 4 for all of my matches in the system. So even when I played matches where TR spit out 5.0 ratings, UTR still has me as a 4 (equivalent to a 3.0 ntrp).

This doesn’t seem to have anything to do with small data set. It seems like it exposes that the assumption that rating should decay with inactivity is very flawed.

I suspect that it assumes that a high 4.5 becomes a 3.0 if he sits out a year of usta.

If it makes you feel any better my friend who played for a ranked D1 and had ATP points was a 5 UTR after a season of 10.0 mixed.

His NTRP was higher than his UTR.

J

 

[S&V-not_dead_yet]

I suspect that it assumes that a high 4.5 becomes a 3.0 if he sits out a year of usta.

My UTR was around 7.5 pre-pandemic and when I resumed > a year later, I was just under 7.

 

[jmnk]

which part of the following statement (written in red letters under your name on UTR site):
" Player metrics are not reliable because player does not have enough match results. "
is unclear??

 

[Moon Shooter]

UTR does not use any results after 12 months at all. This is a huge flaw for this system when it comes to adult rec tennis. This may work well for a high school kid to not have his sophomore results effecting his junior year rating but it is a disaster for adult rec tennis.

I have played 5 mixed doubles matches. The last one was in November 7th. For a few weeks after that last match I had a Provisional rating. But then for some reason my rating lost the "p" "provisional" rating. It can only mean that my opponents or the opponents of my opponents played enough games that UTR algo thinks I have a stable rating now.

So you were treated like a completely new player that only played one match. Your rating based on a single match is not going to be accurate.

UTR constantly readjusts ratings every day. As your opponents change their rating and the opponents of your opponents change ratings I believe your rating changes daily. I do not think your rating necessarily goes down over time. Mine has been going up or down daily. 3 of my matches were with the same partner. I have noticed that on days when my rating is high her rating is low and vice versa. UTR is likely looking at her opponents in games I was not involved in as well as my opponents and calculating that into what it believe is our rating. How far these ripples go I don't know. Because there is a top end to UTR it would seem that every time novak djokovic loses a game everyone's rating goes up.

Because 1)USTA never seems to allow two women to play two men or singles games of men versus women and 2) this is what seems to be the almost exclusive source of UTR games in UTR, the ratings in UTR may suffer in some areas. In my area I believe the women's UTR may be a bit inflated - especially in doubles. I suspect that in my area if two men teamed up and played doubles against two women they would almost certainly increase their doubles utr.

 

[travlerajm]

Anyone trying to blame UTR’s inaccuracies on limited data is in denial about the seriousness of problems with the algorithm.

The algorithm should be able to calculate a single match rating. That match rating may not fully represent a player’s average level, but it should at least make logical sense.

If 3 players on the court have extensive match data. UTR should be able to reasonably estimate the level of the 4th player based on the match result. But the algorithm clearly failed miserably to do that.

The fact is I played a match against a high 4.0C male player and a low 4.5C female player. My partner was a mid 3.5C female. All 3 of these players has verified UTR rating matching their NTRP rating. My partner and I won equal number of games as our opponents. My UTR based on that match should this reflect my 4.5 rating as the performance was still consistent with this, but instead UTR spits out a rating that says I’m a 3.0-3.5 ntrp player, amd then assigns that 3.0-3.5 ntrp rating for my previous matches as well, including those where TR says I played with 5.0 match rating. Does not compute.

 

[DCNJ]

What is the algorithm? What are the issues that could arise in the algorithm and how does it handle those issues?

I could think of a couple plausible things that might cause the algorithm to underestimate the level of someone with very limited match play history--there are going to be errors in the estimation. In many instances the errors are just as problematic if they lead to higher or lower estimates than the actual, and in that case one would assume that the estimate that is produced would be assuming the error is zero. However, there are applications where over- or underestimating would lead to larger costs or problems than the opposite. In that case it's best to make an estimate that would be biased in one direction or the other.

Potentially--it might be those that randomly do better in their first forays into things that can be rated (the UTR events are one of the things I'm thinking about here) may tend to participate more often than those that don't do as well, which would cause a bias and issues when those people, whose ratings are higher than they 'should be'. So underrating a player's first match or two might make those players who have a second, third, or fourth match more accurate.

So maybe in addition to just the small sample size issue causing the estimate to be unreliable (and UTR saying that the estimate is unreliable) there are other issues of which you are unaware that are also covered under the "estimate is unreliable" text.

 

[S&V-not_dead_yet]

Anyone trying to blame UTR’s inaccuracies on limited data is in denial about the seriousness of problems with the algorithm.

The algorithm should be able to calculate a single match rating. That match rating may not fully represent a player’s average level, but it should at least make logical sense.

If 3 players on the court have extensive match data. UTR should be able to reasonably estimate the level of the 4th player based on the match result. But the algorithm clearly failed miserably to do that.

The fact is I played a match against a high 4.0C male player and a low 4.5C female player. My partner was a mid 3.5C female. All 3 of these players has verified UTR rating matching their NTRP rating. My partner and I won equal number of games as our opponents. My UTR based on that match should this reflect my 4.5 rating as the performance was still consistent with this, but instead UTR spits out a rating that says I’m a 3.0-3.5 ntrp player, amd then assigns that 3.0-3.5 ntrp rating for my previous matches as well, including those where TR says I played with 5.0 match rating. Does not compute.

You'll be right if you keep playing matches with results like this and continue to be mis-rated. I suspect as you get more matches under your belt, the discrepancy will decrease. If so, the problem is mainly due to limited data, not a flaw in the algo.

 

[schmke]

You'll be right if you keep playing matches with results like this and continue to be mis-rated. I suspect as you get more matches under your belt, the discrepancy will decrease. If so, the problem is mainly due to limited data, not a flaw in the algo.

The problem is due to both. Yes, more data would help, but you can't give them a pass for doing a poor job with limited data if they could reasonably do better with that data as @travlerajm outlined.

 

[travlerajm]

Played a 2nd usta mixed match today. I felt more like myself, back in my 8.0 groove, and was able to take over the match. Playing with low 3.5 partner, we won easily against another 4.0/4.5 combo pair.

I expect my match rating for today should be close to the 4.5-5.0 edge, which should translate to at least an 8 UTR. Curious to see if UTR will keep me a 4, or adjust.

 

[travlerajm]

Update:

I just checked my UTR. It has gone even lower since yesterday. Still a 4, but now it has me as a lower 4. What’s going on?

 

[Moveforwardalways]

Playing mixed doubles will tank your UTR. This is why almost no adults in my area like UTR and don’t really follow it or give it much credence. This is also why the USTA does not count it as part of your NTRP rating. Combo would have a similar effect, and USTA does not factor in combo either. Interestingly, TR also factors in mixed and combo, which artificially lowers their ratings compared to USTA.

 

[socallefty]

Update:

I just checked my UTR. It has gone even lower since yesterday. Still a 4, but now it has me as a lower 4. What’s going on?

Maybe the UTR officials monitor TTW and they are deliberately screwing with you as they don’t want others to follow your unusual ideas about racquet customization, wearing wristbands to affect SW etc.

 

[travlerajm]

Update:

As of today, my match history from earlier than 12 months has been cleansed from my UTR profile.

The only match showing is the one from last week, where I played at roughly UTR 7 level. And my UTR has been upgraded from a 4 to a 7. And since I played at about 8.5 level yesterday, I assume that my UTR rating will be back up close to 8 in a week when my 2nd match rating gets calculated.

So this part at least makes sense. My current UTR is based on one match now.

What doesn’t make sense is why UTR dropped my rating from an 8 to a 4 for a week after playing my first match in 2 years, but apparently before the new match rating had been calculated. The 4 rating didn’t seem to be based on any actual match result. This is where the hole in the algorithm seems to be — the assumed rating decay with inactivity is a bit exaggerated for adult league players (who drop in and out of league play due to life circumstances, but rarely really change in level much)?

 

[Shaolin]

UTR is bizarre. I play a guy who's a strong 5.0 but says his UTR is 7. Isn't 5.0 around 10 UTR? Their rating system seems all over the map.

 

[travlerajm]

UTR is bizarre. I play a guy who's a strong 5.0 but says his UTR is 7. Isn't 5.0 around 10 UTR? Their rating system seems all over the map.

I think your friend’s UTR example is typical.

I will continue to very critical of UTR until they address the serious and obvious flaws in the algorithm. I’m surprised at so many people defending it, when it clearly fails to do what it says was its mission - to create a rating system that allows all players to be compared.
I will be first to commend UTR after they fix the problems.

I think the rating decay with inactivity is one of several flaws.

Another major flaw is the way that UTR appears to adjust your rating when past opponents improve or decline. This inherently creates a situation where Junior results and adult results are incompatible. Every Junior has a rating that drifts upward when he stops playing (because his past opponents are improving), while every adult has a rating that decays downward when he stops playing (to keep the total average rating for the ecosystem constant, or else the whole system implodes and spirals upward). This is how you end up with adult strong 5.0 ntrp rated 7, while a good high school player might be rated 9. The UTR 7 adult league player would crush the UTR 9 Junior, but since they never play each other, there is no mechanism for the algorithm to correct its inherent flaw. So the bug never gets fixed.

 

[Shaolin]

I think your friend’s UTR example is typical.

I will continue to very critical of UTR until they address the serious and obvious flaws in the algorithm. I’m surprised at so many people defending it, when it clearly fails to do what it says was its mission - to create a rating system that allows all players to be compared.
I will be first to commend UTR after they fix the problems.

I think the rating decay with inactivity is one of several flaws.

Another major flaw is the way that UTR appears to adjust your rating when past opponents improve or decline. This inherently creates a situation where Junior results and adult results are incompatible. Every Junior has a rating that drifts upward when he stops playing (because his past opponents are improving), while every adult has a rating that decays downward when he stops playing (to keep the total average rating for the ecosystem constant, or else the whole system implodes and spirals upward). This is how you end up with adult strong 5.0 ntrp rated 7, while a good high school player might be rated 9. The UTR 7 adult league player would crush the UTR 9 Junior, but since they never play each other, there is no mechanism for the algorithm to correct its inherent flaw. So the bug never gets fixed.

Agree w all of this.. The mission of UTR is great but it's so disorderly and random it seems useless. I really don't like that your rating can go up or down based on your opponents results improving/declining. What if they have an injury, should you be punished for that? Also your comparison is spot on, I know a freshman high school UTR 9 that would be triple bagelled by the 5.0 guy I mentioned.

 

[travlerajm]

Agree w all of this.. The mission of UTR is great but it's so disorderly and random it seems useless. I really don't like that your rating can go up or down based on your opponents results improving/declining. What if they have an injury, should you be punished for that? Also your comparison is spot on, I know a freshman high school UTR 9 that would be triple bagelled by the 5.0 guy I mentioned.

Aging adults get injured all the time, while juniors are virtually indestructible. This amplifies the discrepancy between adult and junior ratings.

 

[RyanRF]

It took a while.. maybe 8 matches or so, but eventually my rating settled in a level that I think is appropriate. When I play against other people with well established ratings I feel their numbers are also correct.

Ratings for limited number of matches can be all over the place. Also if you only have a couple of matches, and those opponents only had a couple of matches, then the whole thing is still up in the air. You could stop playing for a few months and just watch your rating drift around based on how the ratings of your prior opponents change.

But in my experience UTR has been accurate for people that use the system as it is intended: frequently playing matches against opponents who also frequently play matches.

All of this is with singles though. I imagine doubles is more difficult to get accurate.

 

[BallBag]

@travlerajm If you go to your Stats page for doubles, there is chart with result analysis. The opponent team UTR calculation might shed some light on how you got that rating. Looking at mine, I'm seeing a 6.79/6.18 team rated around 7.6 and a 6.38/6.68 team rated around 6.1. I'm guessing that calculation includes an adjustment for your partner because I had a much stronger partner for the second match. So it's like you're rating is playing against all three players on court.

 

[Shaolin]

Aging adults get injured all the time

Don't I know it

 

[travlerajm]

@travlerajm If you go to your Stats page for doubles, there is chart with result analysis. The opponent team UTR calculation might shed some light on how you got that rating. Looking at mine, I'm seeing a 6.79/6.18 team rated around 7.6 and a 6.38/6.68 team rated around 6.1. I'm guessing that calculation includes an adjustment for your partner because I had a much stronger partner for the second match. So it's like your rating is playing against all three players on court.

I just looked at my page, and all it revealed was more of the problem.

My Saturday match now shows up, with all 4 players’ pre-match UTR rating.

Me 4.5C, UTR 7.
Male Opponent 4.0C, UTR 5.
My female partner 3.5C, UTR 4.
Female Opponent 4.5C, UTR 4.

So my UTR will be updated based on these numbers.

Can anyone spot the problem here?

 

[jdawgg]

I used to be like travlerajm and thought UTR was garbage and all over the place. That was before the men's 4.5 season. After the season I left feeling like it was way ahead of TennisRecord/USTA. This was especially evident in mens doubles where I think TennisRecord showed that if a player had bad results early and then played with a strong partner, it would attribute all the rating to the strong partner. From playing with those people I knew they were way underrated by TennisRecord and probably USTA. Yet their UTR was very similar to their partners.

The other phenomenon that I see with USTA/TennisRecord is that when you play against a player far below your level, it always seems to substantially underrate you. I'm not the type to double bagel people all that often, but it's clear those players would never win against me. UTR fixed this by not putting much weight on matches where the gap is greater than 2 UTR, unless the worse player wins.

At the end of the season, when everyone had lots of matches ,UTR really impressed me.

 

[RyanRF]

I just looked at my page, and all it revealed was more of the problem.

My Saturday match now shows up, with all 4 players’ pre-match UTR rating.

Me 4.5C, UTR 7.
Male Opponent 4.0C, UTR 5.
My female partner 3.5C, UTR 4.
Female Opponent 4.5C, UTR 4.

So my UTR will be updated based on these numbers.

Can anyone spot the problem here?

You're starting with the assumption that NTRP is correct, and that UTR needs to align with NTRP in order to prove its accuracy.

Have you considered that maybe it's the NTRP system that's creating the appearance of a problem? NTRP ratings are big overlapping buckets, and you don't get to see the exact NTRP of each player (eg 3.86) ... only the broad group they belong to.

4.5 Player A and 4.0 Player B could be either miles apart in skill (4.49 and 3.51) or nearly identical in skill (4.01 and 3.99).

The only way to know if UTR is 'getting it right' is to see if UTR match results frequently align with match predictions per UTR ratings. So in other words UTR just needs to show that it agrees with itself.

... compare this to NTRP system, which hides its detailed ratings and therefore prevents you from evaluating if their algorithm is any good. Although the fact that they have a ratings appeal process at all indicates they are far from perfect and they know it

 

[jmnk]

Dear statisticians: you may want to take a closer look at your entire probabilistic theory. You say that getting either head or tails on a coin flip is equally likely. And you say that I when I do flip coins I should be getting the same number of heads as tails.

Well, I just flipped the coin 10 times. I got 7 heads and only three tails.

Statisticians, you might want to double-check your theory and make sure your logic does not have a glitch in there.

 

[travlerajm]

A tennis match is not a single data point. A tennis match contains roughly 20 games, with each game representing a “trial” that gives useful data. Within 3 matches, there have been roughly 60 trials, enough to give a reasonably accurate assessment of a player’s rating.

If UTR is accurate after 10+ matches, that is not the point. A large sample size brute-forces an accurate result in spite of flaws in the algorithm that make it less accurate.

If these obvious flaws were to be stripped out, the algorithm wouldn’t need 10+ matches to “settle in.” Three matches would generally be sufficient to know where you stand with a useful and meaningful rating.

TR, TLS, the USTA, and @schmke are using simpler algorithms that don’t have a built-in flaw to cause ratings to implode into a death spiral upon inactivity. And thus these ratings are more meaningful for the vast majority of recreational tennis players, since most players only play a few league or tournament matches per year.

I don’t think I’m asking for much.

 

[S&V-not_dead_yet]

I don't think the algorithm was designed with an 18 month lack of data in mind.

 

[jmnk]

A tennis match is not a single data point. A tennis match contains roughly 20 games, with each game representing a “trial” that gives useful data. Within 3 matches, there have been roughly 60 trials, enough to give a reasonably accurate assessment of a player’s rating.

If UTR is accurate after 10+ matches, that is not the point. A large sample size brute-forces an accurate result in spite of flaws in the algorithm that make it less accurate.

If these obvious flaws were to be stripped out, the algorithm wouldn’t need 10+ matches to “settle in.” Three matches would generally be sufficient to know where you stand with a useful and meaningful rating.

TR, TLS, the USTA, and @schmke are using simpler algorithms that don’t have a built-in flaw to cause ratings to implode into a death spiral upon inactivity. And thus these ratings are more meaningful for the vast majority of recreational tennis players, since most players only play a few league or tournament matches per year.

I don’t think I’m asking for much.

I'm not sure I understand what you are asking for. Judging from your UTR profile you got a pretty accurate UTR estimation after just _two_ matches. Where is the problem then?

 

[Moon Shooter]

Anyone trying to blame UTR’s inaccuracies on limited data is in denial about the seriousness of problems with the algorithm.

The algorithm should be able to calculate a single match rating. That match rating may not fully represent a player’s average level, but it should at least make logical sense.

If 3 players on the court have extensive match data. UTR should be able to reasonably estimate the level of the 4th player based on the match result. But the algorithm clearly failed miserably to do that.

The fact is I played a match against a high 4.0C male player and a low 4.5C female player. My partner was a mid 3.5C female. All 3 of these players has verified UTR rating matching their NTRP rating. My partner and I won equal number of games as our opponents. My UTR based on that match should this reflect my 4.5 rating as the performance was still consistent with this, but instead UTR spits out a rating that says I’m a 3.0-3.5 ntrp player, and then assigns that 3.0-3.5 ntrp rating for my previous matches as well, including those where TR says I played with 5.0 match rating. Does not compute.

The problem is due to both. Yes, more data would help, but you can't give them a pass for doing a poor job with limited data if they could reasonably do better with that data as @travlerajm outlined.

First let me say UTR has many problems. I think the limit on Data is most likely the biggest problem but there are others. Another is the arbitrary cap at the top and bottom. You can cap the bottom but capping the top means the ratings will have to constantly be reshuffling based on the top player doing well or bad and getting closer or further from that cap.

But as far as your view of accurately rating a single game I think as Ryan said you are assuming NTRP has bearing with UTR. It really doesn't. They both work with very different data pools. NTRP includes almost no mixed games unless the person has a mixed exclusive rating. Mixed doubles is a fairly large number of games to use. NTRP goes back 3 years UTR only goes back 12. UTR does not count singles games toward your doubles and vice versa. NTRP does. NTRP is so wide it is almost meaningless. https://cdn.universaltennis.com/public/media/UTR_Player_Range.pdf Consider UTR will not even rate a game with more than a 2 point difference unless the lower rated player wins the entire match. They view such games as completely uncompetitive and therefore it is irrelevant if the lower rated player takes a game or two. Also see how a 4.5 UTR player could be a 3.0 a 3.5 or a 4.0? So even talking about someone's ntrp is getting off on the wrong foot.

So what about focusing just on UTRs? The problem is you think your rating is based on the rating your opponents and your partner had the day before your rating was published. But even if you never played your match their rating would change. After my fifth and final match on November 7th my rating varies from 3.8 to 5.3. This is the case even though I am not playing any other games! How can this be? In part is is because my partners or my opponents or their partners or opponents of my opponents are all playing more games and they are therefore getting shuffled in. They are all also having games dropped off because they are past the magical 12 month deadline. Dropping off those games would not be a big deal if there was more data but not only is it a big problem when there isn't enough data the cut off is itself limiting the data! It really is unbelievably stupid unless you are thinking in terms of a high school kid going from Sophomore to Junior year and you don't want a few bad games as a sophomore to screw up a juniors UTR as he or she is looking to be recruited for college. So since we don't know what their rating would have been if you didn't play your match you don't know how your match effected their rating. You may think they got some sort of boost that day from your match but the boost could be from things that are completely different than your match.

I think this lack of clarity as to how UTR is spitting out numbers does hurt its legitimacy. I think NTRP is much better since people can see how it is done and gain a better understanding. But just because UTR seems to be pulling numbers out of the sky that does not mean eventually with enough matches for you, your partners, your partners other partners and and your opponents and their opponents etc it won't settle in to an accurate rating. It is just that the way they set up the system in some areas it will never be the case that adult rec players will play enough rated games against and with people that play enough rated games that this will happen.

So yes your initial rating makes no sense primarily due to a lack of data.

 

[Goof]

FWIW, my UTR just updated to 6.17 (no idea what it was previously), and I'm currently NTRP 4.0 (I was 4.5 until 12/2019 and have only played 3 league matches since then, three close singles losses against 4.5s).

Edit: forgot one match: a 6-0 6-1 win in a 4.0 singles match against a player who is now a 3.5.

 

[travlerajm]

Update:

My UTR updated after this weekend’s match.

My mixed partner and I received a combined 0.4 UTR bump up. But 0.3 of the bump up was given to my 3.5 partner, and only 0.1 to me.

 

[doublesplayer]

I just played my first USTA match in nearly 2 years.
8.0 mixed.
Yeah. I know I played like crap. I was rusty from not touching racquet in a week. I choked a few points in the super.

But…

After the match, I looked on UTR site, and my UTR has been updated to account for the new data.

My UTR has been downgraded… A lot. We played against a team with combined 11 UTR. Won equal games. My partner was a UTR 4. The other 3 players on court having been playing frequently this year so lots of data on them. Shouldn’t my match rating be a 7 then ?

UTR now says I’m a 4. A UTR 4.xx with 2 decimal place accuracy. Same as my female 3.5 mixed partners. Is it the rating decay with inactivity thing?

TennisRecord says I’m still about 4.4 NTRP (high 4.5) same as I’ve been for the last few years. TLS says I’m a 4.4 too based on my early 2021 results. And the USTA says my computer rating is in range of appealing up to 5.0. And the last year I played a few usta matches, in 2019, I made the TT 4.5 all-star team, which meant schmke had me at least 4.41.

But of course, UTR is the world’s most accurate algorithm. So maybe they know something others don’t?

I mean… I’m ok being a 4. This way my opponents won’t expect much. But UTR, just saying, you might want to double-check your algo and make sure your coder didn’t leave a glitch in there.

1.TR and TLS mean nothing to UTR and USTA
2.Where on USTA does it tell you, you are within range of appealing to 5.0?

 

[travlerajm]

1.TR and TLS mean nothing to UTR and USTA
2.Where on USTA does it tell you, you are within range of appealing to 5.0?

If you do not see the “automated appeal available” button on your usta profile, according to the usta, it means you are not close to the edge of your ntrp rating band (either bottom edge or top).

 

[schmke]

If you do not see the “automated appeal available” button on your usta profile, according to the usta, it means you are not close to the edge of your ntrp rating band (either bottom edge or top).

I'm pretty sure they don't telegraph the auto-appeal response by displaying something different. I believe there is just an appeal link/button for all.

What led you to believe it was shown depending on how close you are?

 

[travlerajm]

I'm pretty sure they don't telegraph the auto-appeal response by displaying something different. I believe there is just an appeal link/button for all.

What led you to believe it was shown depending on how close you are?

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.

 

[esgee48]

Good thing. If you were granted an appeal down, but not the appeal up, where would you be? I swear, if there was a sign saying 'No Fire' in a room full of gunpowder, you would light a match to to see what the fuss was all about.

 

[J011yroger]

Good thing. If you were granted an appeal down, but not the appeal up, where would you be? I swear, if there was a sign saying 'No Fire' in a room full of gunpowder, you would light a match to to see what the fuss was all about.

That's why we love him!

J

 

[schmke]

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.

So, when I'm logged in I see a banner saying "Welcome" and on the right side it has this:


NTRP Level: 4.5
Appeal Rating Level
USTA Account #: XXXXXXXX
Manage Account

This is what has always been there for me and the link hasn't come and gone over time. Are you seeing something different?

 

[travlerajm]

So, when I'm logged in I see a banner saying "Welcome" and on the right side it has this:


NTRP Level: 4.5
Appeal Rating Level
USTA Account #: XXXXXXXX
Manage Account

This is what has always been there for me and the link hasn't come and gone over time. Are you seeing something different?

Yes it was different.

Edit: I also notice my TR.com profile says “appeal available” next to my rating. I assume this designation is auto-imported from the usta site like the rest of their data.

 

[schmke]

Yes it was different.

Edit: I also notice my TR.com profile says “appeal available” next to my rating. I assume this designation is auto-imported from the usta site like the rest of their data.

So, I'm still confused. You are saying it is telling you appeal available because it is letting you know you are in range, but then you tried appealing up and down and were denied. So whatever it is saying isn't an indication you are in range?

Still not sure where you are seeing "automated appeal available" on TennisLink and why you believe it means you are in range.

 

[travlerajm]

So, I'm still confused. You are saying it is telling you appeal available because it is letting you know you are in range, but then you tried appealing up and down and were denied. So whatever it is saying isn't an indication you are in range?

Still not sure where you are seeing "automated appeal available" on TennisLink and why you believe it means you are in range.

After you use the appeal option, the options change. But the “Appeal Available” still appears on the TR site next to my rating because TR hasn’t been updated since I tested the appeal on the usta source site.