is an 11 UTR really a 4.5 NTRP?

[LOBALOT]

There's a big difference between "rarely" and "never". These matchups are rare but they do happen, and it probably doesn't take a huge number of them to be able to set a reasonable standardization.

USTA faces a similar issue with standardizing rec-level NTRP across all the regions of the U.S. They have to rely on results from National tournaments, and perhaps players who move from one region to another. Those are are only a tiny fraction of all matches played, yet they carry large weight in determining ratings for eveyone, and it seems to be reasonably successful, though some probably disagree.

I would think that the UTR system somehow applies extra weight to results from rare matchups that link different parts of the network, though I don't know for sure how well they accomplish that.

Unfortunately I don't think they do if you look at their web site they tell you their algorithm and really other than recent matches having more weight they don't indicate they take other factors into account.

 

[schmke]

There's a big difference between "rarely" and "never". These matchups are rare but they do happen, and it probably doesn't take a huge number of them to be able to set a reasonable standardization.

USTA faces a similar issue with standardizing rec-level NTRP across all the regions of the U.S. They have to rely on results from National tournaments, and perhaps players who move from one region to another. Those are are only a tiny fraction of all matches played, yet they carry large weight in determining ratings for eveyone, and it seems to be reasonably successful, though some probably disagree.

I would think that the UTR system somehow applies extra weight to results from rare matchups that link different parts of the network, though I don't know for sure how well they accomplish that.

Good point that UTR may very well have results that tie the different areas or cohorts together, but they would have to have a more sophisticated way of identifying and using those matches than the USTA does. The USTA has it easy in that the important matches they need to use are self-identifying, e.g. Nationals, Sectionals, States/Districts, so their year-end algorithm can specifically use these. UTR doesn't necessarily have these clearly identified as an individual from another area/country could end up playing anywhere.

I'm not suggesting UTR doesnt' attempt to do something like this, I don't know either way, but if they do, that might explain some of the volatility of their rating if they try to factor this in too aggressively.

 

[J011yroger]

If adults rarely play juniors, women rarely play men, players in different countries don’t play each at the rec level etc. (particularly in singles), how does UTR know how to standardize all levels of groups playing across the whole world? Should we just believe that somehow UTR knows what they are doing?

Bunch of my kids went to South America to play ITFs. They all came back with higher UTRs. They all said most of the players were .5-1 overrated compared to NY.

J

 

[travlerajm]

Bunch of my kids went to South America to play ITFs. They all came back with higher UTRs. They all said most of the players were .5-1 overrated compared to NY.

J

I can confirm that 90% of teaching pros in South America have never heard of UTR and don’t care.

 

[TennisOTM]

Unfortunately I don't think they do if you look at their web site they tell you their algorithm and really other than recent matches having more weight they don't indicate they take other factors into account.

You might be right, or it could be that they don't mention it on the website because it's not a routine part of the algorithm but something that is occasionally done globally as a sort of system adjustment.

There's also some sense in which these rare "cohort-linking" match results can become sort of naturally critical to the system without having to explicitly identify them and give them more weight.

Imagine there are two cohorts of players, each with 100 members. The players play lots of matches within their cohort but there are never any matches between players from different cohorts. In this example, there is no way to tell whether one cohort is better than the other on average. But then say there is one player introduced who has had 10 opponents from one cohort and also 10 opponents from the other.

As the system iterates through match results and ratings calculations, those 20 match results in theory should naturally serve to compare the cohorts, even without explicitly telling the system to give them more weight. If the cohorts are very different then it might take awhile for the effect to filter through. Would be interesting to hear from UTR on how this kind of thing works in practice.

 

[travlerajm]

You might be right, or it could be that they don't mention it on the website because it's not a routine part of the algorithm but something that is occasionally done globally as a sort of system adjustment.

There's also some sense in which these rare "cohort-linking" match results can become sort of naturally critical to the system without having to explicitly identify them and give them more weight.

Imagine there are two cohorts of players, each with 100 members. The players play lots of matches within their cohort but there are never any matches between players from different cohorts. In this example, there is no way to tell whether one cohort is better than the other on average. But then say there is one player introduced who has had 10 opponents from one cohort and also 10 opponents from the other.

As the system iterates through match results and ratings calculations, those 20 match results in theory should naturally serve to compare the cohorts, even without explicitly telling the system to give them more weight. If the cohorts are very different then it might take awhile for the effect to filter through. Would be interesting to hear from UTR on how this kind of thing works in practice.

Consider this scenario: an adult player from the US who normally practices about 1h per week at home against a wall, plus playing occasional usta mixed doubles and occasional UTR flex singles without any singles practice between matches. UTR assigns him a rating.

Then that player gets stationed in South America on project for a year, away from family, so all he has to keep him busy is playing singles tennis 4-5x per week. His level of course goes up by a few UTR points compared to his level at home. He enters a few UTR singles tourneys down there and records matches. What would that do to the UTR system?

 

[jhick]

Bunch of my kids went to South America to play ITFs. They all came back with higher UTRs. They all said most of the players were .5-1 overrated compared to NY.

J

So if you want to increase your UTR, travel and play in tournaments in South America.

 

[J011yroger]

So if you want to increase your UTR, travel and play in tournaments in South America.

Your tactics confuse and frighten me sir.

J

 

[TennisOTM]

Consider this scenario: an adult player from the US who normally practices about 1h per week at home against a wall, plus playing occasional usta mixed doubles and occasional UTR flex singles without any singles practice between matches. UTR assigns him a rating.

Then that player gets stationed in South America on project for a year, away from family, so all he has to keep him busy is playing singles tennis 4-5x per week. His level of course goes up by a few UTR points compared to his level at home. He enters a few UTR singles tourneys down there and records matches. What would that do to the UTR system?

I smell sarcasm, but your own example probably does exemplify a tough conundrum for UTR. On the one hand they should be making good use of the rare match results that link groups of players who don't routinely play each other. On the other hand, many of those matches are probably played in low-stakes or casual scenarios where it's unclear how much the data can be relied upon as a true benchmark.

Male vs. female ratings is another example. While I'm sure UTR has some data on male-v-female matchplay in singles, how many of those matches occur in competitive scenarios where we have high trust that the result/score reflects true skills of each player?

 

[travlerajm]

I smell sarcasm, but your own example probably does exemplify a tough conundrum for UTR. On the one hand they should be making good use of the rare match results that link groups of players who don't routinely play each other. On the other hand, many of those matches are probably played in low-stakes or casual scenarios where it's unclear how much the data can be relied upon as a true benchmark.

Male vs. female ratings is another example. While I'm sure UTR has some data on male-v-female matchplay in singles, how many of those matches occur in competitive scenarios where we have high trust that the result/score reflects true skills of each player?

my doubles UTR went up 2 UTR units overnight. Yesterday I was a 6. Today I’m an 8. I haven’t played a match in a month. but I have 7 doubles matches logged in their system this year. Until UTR owns up that their algorithm is trash and fixes the errors in the code, I can’t really take UTR as seriously as I do the other rating systems.

 

[FuzzyYellowBalls]

my doubles UTR went up 2 UTR units overnight. Yesterday I was a 6. Today I’m an 8. I haven’t played a match in a month. but I have 7 doubles matches logged in their system this year. Until UTR owns up that their algorithm is trash and fixes the errors in the code, I can’t really take UTR as seriously as I do the other rating systems.

4% reliability on that 8, your 100% 6 for singles is probably the one you should focus on, and that is accurate.

 

[J_R_B]

I think the list below works pretty well as a rule of thumb for men. Obviously we have all seen exceptions but I think it's pretty accurate in most cases where the player has good data (many matches with a good network of opponents).

NTRP 2.5 = UTR <2.25
NTRP 3.0 = UTR 2.25-3.75
NTRP 3.5 = UTR 3.75-5.25
NTRP 4.0 = UTR 5.25-6.75
NTRP 4.5 = UTR 6.75-8.25
NTRP 5.0 = UTR 8.25-9.75
NTRP 5.5 = UTR 9.75-11.25
NTRP 6.0 = UTR 11.25-12.75
NTRP 6.5 = UTR 12.75-14.25
NTRP 7.0 = UTR 14.25-15.75 (touring pros)
NTRP 7.5 = UTR >15.75 (best pros in the world)

I extended to the highest levels just for fun. Imagine a top-10 ATP player getting upset because he got bumped up to 7.5 and can't play in the 7.0 league anymore.

Do others agree with this? This feels to me like the most believable simplification of the conversion I’ve seen that aligns to my experiences.

It also manages to explain the phenomena i find when I say I’m looking for 4.5-5.0 players (which to me implies ntrp) and then I hit the court with what feels more like a 4.5-5.0 UTR.

I agree with it except that it ignores natural overlap between levels. In other words, I'd call NTRP 4.0 UTR 5.00-7.00 and NTRP 4.5 6.50-9.50. The 6.5-7.0 area is overlap where people are going to bump up and down between the 4.0 and 4.5 NTRP levels every couple years.

 

[MGArchitect]

my doubles UTR went up 2 UTR units overnight. Yesterday I was a 6. Today I’m an 8. I haven’t played a match in a month. but I have 7 doubles matches logged in their system this year. Until UTR owns up that their algorithm is trash and fixes the errors in the code, I can’t really take UTR as seriously as I do the other rating systems.

For doubles, you've played 7 matches of 8.0 mixed doubles as a 4.5 with a 3.5 female partner and have a 4% reliable rating. I am not surprised UTR doesn't know what to do with you, your complaints notwithstanding. You are likely an edge case, exception, etc. Did you play any doubles while in South America that would help normalize regions? For singles, you show a grand total of 4 matches in South America, one of which was a default against you and the other you retired after the first set. One was a double bagel win over someone with a UTR showing 1.xx-3.xx. One double breadstick loss to a current NAIA college player, though looks to be a reasonably good player. These all suggest that there is not enough information to reliably rate you, or for anyone at UTR to take you seriously when you complain. Their algorithm talks about taking the 30 most recent results in the last 12 months. Do you have any 12 month period where you have 30 results? That's who their rating system is meant to measure - juniors, college players, week in and week out league players.

 

[travlerajm]

For doubles, you've played 7 matches of 8.0 mixed doubles as a 4.5 with a 3.5 female partner and have a 4% reliable rating. I am not surprised UTR doesn't know what to do with you, your complaints notwithstanding. You are likely an edge case, exception, etc. Did you play any doubles while in South America that would help normalize regions? For singles, you show a grand total of 4 matches in South America, one of which was a default against you and the other you retired after the first set. One was a double bagel win over someone with a UTR showing 1.xx-3.xx. One double breadstick loss to a current NAIA college player, though looks to be a reasonably good player. These all suggest that there is not enough information to reliably rate you, or for anyone at UTR to take you seriously when you complain. Their algorithm talks about taking the 30 most recent results in the last 12 months. Do you have any 12 month period where you have 30 results? That's who their rating system is meant to measure - juniors, college players, week in and week out league players.

I also won an open level doubles tournament last year.

 

[FuzzyYellowBalls]

For doubles, you've played 7 matches of 8.0 mixed doubles as a 4.5 with a 3.5 female partner and have a 4% reliable rating. I am not surprised UTR doesn't know what to do with you, your complaints notwithstanding. You are likely an edge case, exception, etc. Did you play any doubles while in South America that would help normalize regions? For singles, you show a grand total of 4 matches in South America, one of which was a default against you and the other you retired after the first set. One was a double bagel win over someone with a UTR showing 1.xx-3.xx. One double breadstick loss to a current NAIA college player, though looks to be a reasonably good player. These all suggest that there is not enough information to reliably rate you, or for anyone at UTR to take you seriously when you complain. Their algorithm talks about taking the 30 most recent results in the last 12 months. Do you have any 12 month period where you have 30 results? That's who their rating system is meant to measure - juniors, college players, week in and week out league players.

Good solid details. And you are 100% correct about who the rating system is meant to measure, I still think the irony is you paid UTR for a subscription @travlerajm , not a lot of money, but why even do that?

 

[FuzzyYellowBalls]

I also won an open level doubles tournament last year.

You probably did, but of course open to anyone, no need to provide details.

 

[travlerajm]

Good solid details. And you are 100% correct about who the rating system is meant to measure, I still think the irony is you paid UTR for a subscription @travlerajm , not a lot of money, but why even do that?

Full disclosure is that I paid for the year of UTR power as a scouting tool for my pro clients. My conclusion was that it was pretty worthless.

 

[TennisOTM]

Full disclosure is that I paid for the year of UTR power as a scouting tool for my pro clients. My conclusion was that it was pretty worthless.

Did it not pay for itself with discounts for UTR flex league fees?

 

[travlerajm]

Did it not pay for itself with discounts for UTR flex league fees?

I was already no longer paying for power when I started UTR flex league. They gave me free entry into singles flex leagues for being UTR 7+.

 

[Moon Shooter]

I think the list below works pretty well as a rule of thumb for men. Obviously we have all seen exceptions but I think it's pretty accurate in most cases where the player has good data (many matches with a good network of opponents).

NTRP 2.5 = UTR <2.25
NTRP 3.0 = UTR 2.25-3.75
NTRP 3.5 = UTR 3.75-5.25
NTRP 4.0 = UTR 5.25-6.75
NTRP 4.5 = UTR 6.75-8.25
NTRP 5.0 = UTR 8.25-9.75
NTRP 5.5 = UTR 9.75-11.25
NTRP 6.0 = UTR 11.25-12.75
NTRP 6.5 = UTR 12.75-14.25
NTRP 7.0 = UTR 14.25-15.75 (touring pros)
NTRP 7.5 = UTR >15.75 (best pros in the world)

I extended to the highest levels just for fun. Imagine a top-10 ATP player getting upset because he got bumped up to 7.5 and can't play in the 7.0 league anymore.

I think this minimizes the huge overlap in actual usta ratings. That said if I had to pick hard cut offs these seem reasonable for my area - at least at the lower end that I’m more familiar with. But I don’t know any 3.0 male players with a 2.xx or lower reliable utr. Even the one usta 2.5 guy I know (that plays mixed doubles) is a 3.xx utr player. About 15 of my 3.0 teammates and opponents are/were 4.xx or higher. But more then half of them got bumped up to 3.5.

 

[jmnk]

This is 100% true.

And the reason is obvious. UTR has to maintain its zero sum system.

in what sense is UTR a 'zero sum system'? It simply is not. Are you suggesting that when a new player registers and gets himself let's say UTR 2.0 rating then that 2.0 points has to be deducted from all the players? UTR algorithm does not care if there are 1000 UTR 10.0 players or 2000 UTR 10.0 players.

When a junior takes a 6-month break from competition, his ex opponents improve a lot, and since the algo adjusts your rating if past opponents improve, UTR bumps up the rating of the inactive junior even if the player didn’t actually improve.

perhaps. And then when he comes back after 6 months and starts playing below his now-current level his UTR will get adjusted down. So what's your point?

In order to maintain the zero sum across the whole system, it has to continuously decay all the adult ratings.

really? So you are saying that UTR formula has some secret algorithm that goes over rankings of all the players, and selects just 'adults' to decay their rating. That's quite an assumption.

So the only adults with accurate ratings are the ones who’ve been playing lots of recent matches,

whoa, and that is somehow bad??

and even those ratings are depressed compared to the juniors and college players.

Because again of that secret algorithm that is somehow programmed to punish 'adults only'?

Whenever I play singles against a “100% reliable” junior or college player, i seem to “overperform” because they have inflated rating compared to the adults.

and when have you played a 100% reliable junior or college player in singles?? Because it looks like the last time it happened the result was a 1 and 1 loss, which is pretty much in line with what was expected given yours and your opponent UTR ranking.

 

[jmnk]

This is an excellent point and I am not sure any "comparative" measuring system can get around this issue.

College Tennis is another area where you have to look very carefully at UTRs as some player my join a team as a freshman as an 11 UTR and be spotted at 6 singles in the lineup and play 10s for a year or two and thus their UTR will drift toward that10. They still play at that 11 level but they are only playing 10s so their UTR will be a 10.

why would that be? If they are playing UTR 10s and they win majority of matches (as they should given that they are UTR 11s) - why would their ranking not stay at 11?

I mean sure, it is harder to maintain one's rating playing constantly players below one's level - but that is just the nature of the algorithm based on Elo. Gukesh just won a match becoming World Chess Champion and his ranking _declined_ because he only barely beat Ding who had FIDE ranking like 60 points less (because per Elo Gukesh was supposed to win more convincingly). Does that mean that FIDE Elo ranking is wrong - absolutely not.

 

[jmnk]

For doubles, you've played 7 matches of 8.0 mixed doubles as a 4.5 with a 3.5 female partner and have a 4% reliable rating. I am not surprised UTR doesn't know what to do with you, your complaints notwithstanding. You are likely an edge case, exception, etc. Did you play any doubles while in South America that would help normalize regions? For singles, you show a grand total of 4 matches in South America, one of which was a default against you and the other you retired after the first set. One was a double bagel win over someone with a UTR showing 1.xx-3.xx. One double breadstick loss to a current NAIA college player, though looks to be a reasonably good player. These all suggest that there is not enough information to reliably rate you, or for anyone at UTR to take you seriously when you complain. Their algorithm talks about taking the 30 most recent results in the last 12 months. Do you have any 12 month period where you have 30 results? That's who their rating system is meant to measure - juniors, college players, week in and week out league players.

it is simply unfortunate that you decided to spoil a nice story with facts. We don't do that here

 

[LOBALOT]

why would that be? If they are playing UTR 10s and they win majority of matches (as they should given that they are UTR 11s) - why would their ranking not stay at 11?

I mean sure, it is harder to maintain one's rating playing constantly players below one's level - but that is just the nature of the algorithm based on Elo. Gukesh just won a match becoming World Chess Champion and his ranking _declined_ because he only barely beat Ding who had FIDE ranking like 60 points less (because per Elo Gukesh was supposed to win more convincingly). Does that mean that FIDE Elo ranking is wrong - absolutely not.

As you progress through time your more recent matches carry more weight and your older matches fall off the chart. Those matches that you crushed an 11 a year and a half ago fall of the chart and you are left with matches where you crushed a 10 (As an example) if that is all you play in the range of players at the line you are in the lineup.

They have to take recent matches into account and discard old ones. Players don't progress upward in a linear fashion forever. Moreover, as players age (say hit 35) their level starts to decrease. They aren't going to be an 11 forever.

 

[Roforot]

For doubles, you've played 7 matches of 8.0 mixed doubles as a 4.5 with a 3.5 female partner and have a 4% reliable rating. I am not surprised UTR doesn't know what to do with you, your complaints notwithstanding. You are likely an edge case, exception, etc. Did you play any doubles while in South America that would help normalize regions? For singles, you show a grand total of 4 matches in South America, one of which was a default against you and the other you retired after the first set. One was a double bagel win over someone with a UTR showing 1.xx-3.xx. One double breadstick loss to a current NAIA college player, though looks to be a reasonably good player. These all suggest that there is not enough information to reliably rate you, or for anyone at UTR to take you seriously when you complain. Their algorithm talks about taking the 30 most recent results in the last 12 months. Do you have any 12 month period where you have 30 results? That's who their rating system is meant to measure - juniors, college players, week in and week out league players.

It may be fine for juniors and college players. Though there seem to be ways for game the system. It seems that it is not useful or applicable to rec league players

 

[MGArchitect]

It may be fine for juniors and college players. Though there seem to be ways for game the system. It seems that it is not useful or applicable to rec league players

to be clear, I mean people who get 30 results or more in a 12 month period. That leaves out a large majority of rec league players in my experience. I think the cohort issue definitely is apparent in UTR as there is very little crossover between juniors/college players and USTA league players except for the few who also play UTR tournaments or against 18 year olds in men's open, which is a very limited number, and something @travlerajm actually did do. When I played in college, I played zero outside tournaments or leagues, and I think that is still the case today with most current college players. When I was a junior, I played in USTA tournaments, but only juniors, and only against adults in a local ladder league that even today would never get reported to UTR.

 

[ryohazuki222]

I’ve recently decided that if someone is 7.5-8.5 UTR they are at least 4.5 and not yet 5.5 ntrp.

(Used this thread plus my own recent experiences.)

Per original title to thread… I’d say a 4.5 at 11 UTR is a strong sandbagger or anomaly/error within the rating system.

 

[CosmosMpower]

IMO no way. I think you're closer

My club pro says I'm a 4.0-4.5, and my UTR has never been higher than 5.2

I'm underachieving for sure due to a 2-5 record in 2023 but I think the two numbers are closer than most people believe.

My ex-buddy was a 5.0 for a long time, is probably a 4.5 now after a serious injury. His UTR is 6.XX

I hit with @MyFearHand sometimes. He's a 5.0 and his current UTR looks to be 7.XX although I'm aware that it was around 10 at some point.

But I think at a 10 UTR you aren't a 5.0, more like a 5.5. I've seen 11+ UTR players at the tourneys I play in and they are active D1 players, usually at George Mason University

You’re a middle of the road 3.5

 

[asdad1109]

There is a coach at my tennis club who is a 13 UTR. I was asking him about what the equivalent UTR to NTRP is for 4.5. I asked him if the low end of 4.5 is around a 7 UTR, and he said no way. He said the average 4.5 UTR is a 10, and the upper end is 11 UTR.


Is this true? Are NTRP 4.5's really in double digits?

Bahahahaha 4.5 is like a 6 UTR

 

[Rosstour]

You’re a middle of the road 3.5

I agree completely, thank you

 

[Jono123]

The problem is 4.5 and 5.0 covers quite a broad upper ability span,. Many competitive rec players never break into 5.0

 

[ryohazuki222]

Where it’s really interesting IME though is a good chunk of players are a good 1-2 points better in practice than in competitive matches.

Especially at 4.5-5.0 range.

Ie; a lot of the higher level players there are just much better at actually winning. Play better under pressure, more practiced in competition, etc.

And also the unbalanced players that still suck at serving at that level but have a great groundstroke game. Ie; play like a 5.0 in ground stroke games; but play like a 4.0 in match scenarios because serve plus 1s are garbage and can’t make it to the rally.

 

[GrassCourtFan]

Around here the top NTRP 4.5s are UTR 8, the bottom ones are UTR 6. UTR7 is typical. UTR 10 would be a good 5.0 around here.

 

[J011yroger]

Where it’s really interesting IME though is a good chunk of players are a good 1-2 points better in practice than in competitive matches.

Especially at 4.5-5.0 range.

Ie; a lot of the higher level players there are just much better at actually winning. Play better under pressure, more practiced in competition, etc.

And also the unbalanced players that still suck at serving at that level but have a great groundstroke game. Ie; play like a 5.0 in ground stroke games; but play like a 4.0 in match scenarios because serve plus 1s are garbage and can’t make it to the rally.

It will be a cold day in hell when I lose a groundstroke game to a 4.0.

J

 

[TennisManiac]

Mark's moving down to 4.5??

No way was he ever a 5.0

 

[TennisManiac]

Around here the top NTRP 4.5s are UTR 8, the bottom ones are UTR 6. UTR7 is typical. UTR 10 would be a good 5.0 around here.

Exactly the same around here.

 

[ryohazuki222]

It will be a cold day in hell when I lose a groundstroke game to a 4.0.

J

ya… lose the ground stroke game. But my point is that in this situation the ground stroke game ends with something resembling a less embarrassing score than the match.

Ie; maybe 11-4 in a ground stroke game with some decent rallies in the process. As opposed to 6-0 in a set that barely had real rallies.