Share your WTN - Crowd sourcing the NTRP to WTN mapping
- Thread starter schmke
- Start date
UTR is still the best product so far (it consolidates USTA and UTR results every week and for all matches) except that you have to pay $95 per year to get the 2 decimals for your opponent's rating.
I hope that WTN has just a temporary data issue otherwise it will be totally useless (imagine this mess for the 100 countries which have adopted this global rating...) and they would have missed a great chance to compete against the UTR system. To me this data issue is pretty basic so I am not sure it's a great first signal to launch this important project in that manner...
I hope that WTN has just a temporary data issue otherwise it will be totally useless (imagine this mess for the 100 countries which have adopted this global rating...) and they would have missed a great chance to compete against the UTR system. To me this data issue is pretty basic so I am not sure it's a great first signal to launch this important project in that manner...
You answered your own question already.
ChaelAZ
G.O.A.T.
Honestly, unless you are play higher levels a UTR step doesn't seem that vastly different or worse/better to me. At 4.0 I play (I am a UTR 5.7 singles and 5.2 dubs) I have matches against UTR 4, 5, and 6 and have won/lost to all depending on the day. That isn't worth the $95 for me.
But curious if other find that much variation in UTR steps.
Or maybe that would be a good separate thread.
Moon Shooter
Hall of Fame
I see what mistake I made. My singles level was about midway so I thought the upper range 19.2 was my doubles and my lower range 22.9 was my singles. The people that I thought had no rating just had no singles rating. I see they do have a rating in doubles and the rating I posted for doubles is wrong. Doh!
Have people's WTN ratings been changing even when they haven't played?
BallBag
Professional
You should be able to see your own rating to two decimal places for free. You have to pay to creep on other people though.
Unlike UTR, WTN doesn't change when you don't play. Your rating deviation increases meaning your rating becomes less reliable as time goes on without new results. If rating deviation is tied to Game Zone then your game zone should increase but that's just me speculating.
Moon Shooter
Hall of Fame
Keep in mind that they only use USTA data which often doesn't include many singles matches. Did the singles have a blue checkmark?
Moon Shooter
Hall of Fame
That's great news. That means they can track your rating over time and hopefully you can set a goal to hit a certain rating and it won't just be that you hit that goal for a day and then it vanishes in the ether like UTR.
I have to say this seems like a very nice rating and overall system USTA is putting together. The people I have seen with blue check ratings all seem to have ratings that seem in line with what I know. But sure some of the non-blue check ratings are way out. I wonder if some people are not getting a blue checks because of concerns about names.
But what can make this system great is when it is combined with the tools to easily set up tournaments seeding and pairings etc. If any USTA member can set up a tournament that way (and include people in the tournament that are not paid USTA members) this will be great. USTA could get the $10 entry fee or whatever and that should more than cover the cost of providing the rating and back end stuff.
Edit: I think we should ask if the WTN is blue checked or not.
Edit2: I think women tend to be rated better than they are or men are being rated worse then they are whatever. I think once we have more coed events this rating will even out.
Last edited:
I've been looking deeper into some of the outlier WTNd players to get more specific evidence of algorithm flaws or errors. I mentioned one guy earlier who is a 4.0C and UTR 6.XX doubles and yet somehow has a WTNd of 7.1 (blue checked). He has played a lot of doubles with several different partners, so I looked up the WTN of his recent partners and opponents to see if the outcomes make sense.
As you can imagine, with a WTN of 7.1, with just about any partner his doubles team is going to be favored (according to average WTN) against nearly all opposing teams playing as a 4.0, and often heavily favored. Does this match up with the actual outcomes?
I found 47 sets he played so far in 2022 in which all four players have the blue check for doubles. His team was the underdog by WTN in just 3 (6%) of those sets, yet his team lost 15 (32%) of the 47 sets outright, and lost 42% of the games in those sets.
There were 23 sets in which his team was favored by more than 7 WTN units (average of him and his partner more than 7 better than opponents' average). According to the GameZone ranges, a difference of 7 should be an extreme mismatch, i.e. a bagel with high probability. Yet they lost 3 (13%) of those sets outright and lost 34% of the games played in those sets.
Yes this is just one example, but he was not too hard to find just browsing around players in my circle, so I imagine there are quite a lot of examples where the predictive value of WTN appears to be quite poor, especially for doubles.
As you can imagine, with a WTN of 7.1, with just about any partner his doubles team is going to be favored (according to average WTN) against nearly all opposing teams playing as a 4.0, and often heavily favored. Does this match up with the actual outcomes?
I found 47 sets he played so far in 2022 in which all four players have the blue check for doubles. His team was the underdog by WTN in just 3 (6%) of those sets, yet his team lost 15 (32%) of the 47 sets outright, and lost 42% of the games in those sets.
There were 23 sets in which his team was favored by more than 7 WTN units (average of him and his partner more than 7 better than opponents' average). According to the GameZone ranges, a difference of 7 should be an extreme mismatch, i.e. a bagel with high probability. Yet they lost 3 (13%) of those sets outright and lost 34% of the games played in those sets.
Yes this is just one example, but he was not too hard to find just browsing around players in my circle, so I imagine there are quite a lot of examples where the predictive value of WTN appears to be quite poor, especially for doubles.
schmke
Legend
Nice work.
Yes, doubles seems worse than singles, but even singles has some pretty broad ranges.
Moon Shooter
Hall of Fame
Is the guy playing mixed doubles? I think it is pretty clear that women of the same usta dynamic rating are assumed to be the same strength or even stronger than the men at that same rating.
Once we start having men and women play singles against each other and we start having doubles matches with MM v FF or FM v mm and FM v ff these issues should get sorted out fairly quickly. I can see that if I wanted to dramatically boost my singles rating I could play some women and get a very substantial boost. The same goes with teaming up with some guys I know and playing against some women pairs in doubles.
He does play mixed, but I looked at plenty of other guys who play mixed and many of them have WTNd numbers on the opposite side of the distribution. This also doesn't explain why his match outcomes were so misaligned with the WTNd numbers of the four players.
schmke
Legend
Wow, there are not many (less than 1%) with the same WTNs and WTNd.
Cindysphinx
G.O.A.T.
3.5 female
25.4 doubles
Feels about right. It has been a tough couple of years. . . .
25.4 doubles
Feels about right. It has been a tough couple of years. . . .
I found that interesting too especially since I play at least twice as many doubles matches than singles matches
Didn't seem like WTN updated today...did anyone see otherwise?
schmke
Legend
Yep, saw updates as of this morning.
schmke
Legend
Regardless, that is a remarkably large difference. There are 3.0 and 3.5 men and women with better doubles WTN. Makes you go hmmm.
ok, I guess there is a little movement for guys in my area. My numbers didn't move, and they definitely should have (e.g., went from two total singles matches to three).
schmke
Legend
I don't think they got your matches. When were they played/entered?
haha, very consistent indeed!
The WTN is supposed to be updated every Wednesday as stated on their website: "The World Tennis Number will update every Wednesday taking results up to the previous Sunday into account. It will provide a 'real time' standard that allows a player to more closely track their competition performance and progress over time."
OnTheLine
Hall of Fame
From 6/2 - 6/14 ... so the 14th shouldn't be there but the other 7 probably should
I took a look at another questionable WTNd for a 4.0C player, this time on the weaker end of the distribution: a blue-check WTN 25.8d with many recent doubles matches. His W/L record this year is not good (6-16), but his match scores have been good enough to keep him above 3.6 on Tennisrecord above 5.0 for UTR doubles. So why is his WTNd so poor?
I pulled the WTNd for his partners and opponents for 2022 doubles matches and kept the ones with blue checks for all 4 players. In these matches, he and his partner were favored according to averaged WTNd in just 6 of the 40 sets (15%), yet they won 11 (28%) of the sets outright and 44% of the games in those sets.
There were 22 sets in which he and his partner were massive underdogs according to WTN, by amounts ranging from 4.9 to 10.2 (difference in average partnership WTNd). Indeed he lost all 22 of those sets, but he won 37% of the games and none of the 22 sets were 6-0.
That last observation makes me wonder if the WTN really does ignore the set score in their algorithm. If the algorithm only knows that he was 0-22 in sets against those opponents, that would explain why it allowed the huge differences in WTN to persist.
However if the magnitude of WTN difference is supposed to be predictive of the likelihood of a blowout or of the competitiveness / compatibility of a matchup, it seems to be failing in that goal for this player and many others.
I pulled the WTNd for his partners and opponents for 2022 doubles matches and kept the ones with blue checks for all 4 players. In these matches, he and his partner were favored according to averaged WTNd in just 6 of the 40 sets (15%), yet they won 11 (28%) of the sets outright and 44% of the games in those sets.
There were 22 sets in which he and his partner were massive underdogs according to WTN, by amounts ranging from 4.9 to 10.2 (difference in average partnership WTNd). Indeed he lost all 22 of those sets, but he won 37% of the games and none of the 22 sets were 6-0.
That last observation makes me wonder if the WTN really does ignore the set score in their algorithm. If the algorithm only knows that he was 0-22 in sets against those opponents, that would explain why it allowed the huge differences in WTN to persist.
However if the magnitude of WTN difference is supposed to be predictive of the likelihood of a blowout or of the competitiveness / compatibility of a matchup, it seems to be failing in that goal for this player and many others.
FuzzyYellowBalls
Legend
Neither of mine changed either. Maybe, there is something to feeding the WTN enough data so it feels comfortable in the number, I have plenty of matches singles and doubles for calculation, steadily on a weekly basis.
This FAQ from WTN themselves will help focus the comments here, I think everyone should read it before guessing about things: https://worldtennisnumber.com/eng/faq
Quick takes that are detailed: They go back 4 years. The Game Zone range is something they mention as what to look at if you signed up at your local club or park for a match and wanted to find someone good to play against, match up the Game Zone ranges. Not a lot of mention of tournaments, if at all in first glance.
Different goals than UTR, also than USTA in my opinion in general, although cross-gender cross-age, and global
is similar to UTR.
NastyWinners
Hall of Fame
NTRP: 4.5C
WTN Singles: 11.3
WTN Doubles: 10.1
WTN Singles: 11.3
WTN Doubles: 10.1
schmke
Legend
This is certainly what they say, from the USTA FAQ:
But in practice we've seen player's WTNs changing with no apparent additional matches. Perhaps this is just early bugs/process be worked out, but I'm not sure WTNs really won't change.
BallBag
Professional
Maybe we'll have some more information when thier website goes live. Should at least be able to track changes.
schmke
Legend
If you check your USTA profile every Wednesday, you can track your changes too of course. I'm not sure when the ratings are going to show up on the ITF's WTN site but I read something somewhere that it was supposed to be launched and provide more detail than what is on your USTA profile.
schmke
Legend
I looked into players at the extremes a bit and learned some more.
At the larger number end, I do see a few NTRP 2.5s that are 40.0 WTNs. Not a lot, but a few. I doubt they are exactly 40.0 so they seem to be artificially capping WTNs at 40.0. Note some of these player do have the blue check meaning high confidence.
On the lower number end, I only see a few players in the 1.x range, but interestingly if I find a pro, say Jack Sock, there is no WTN shown, and instead he is simply "Pro Zone". It appears anyone with an ATP or WTA ranking is automatically a "pro" and at least the USTA profile doesn't show the WTN. I don't know if this means it isn't even calculated though. I'd think it is calculated, but perhaps it is not artificially limited to 1.0 and so it would look goofy to show it.
Given where we see some recreational league players WTNs in the single digits, it would seem there is not room for pros to fit, and the approach take is to just call them pros and not publish their WTN.
At the larger number end, I do see a few NTRP 2.5s that are 40.0 WTNs. Not a lot, but a few. I doubt they are exactly 40.0 so they seem to be artificially capping WTNs at 40.0. Note some of these player do have the blue check meaning high confidence.
On the lower number end, I only see a few players in the 1.x range, but interestingly if I find a pro, say Jack Sock, there is no WTN shown, and instead he is simply "Pro Zone". It appears anyone with an ATP or WTA ranking is automatically a "pro" and at least the USTA profile doesn't show the WTN. I don't know if this means it isn't even calculated though. I'd think it is calculated, but perhaps it is not artificially limited to 1.0 and so it would look goofy to show it.
Given where we see some recreational league players WTNs in the single digits, it would seem there is not room for pros to fit, and the approach take is to just call them pros and not publish their WTN.
BallBag
Professional
That's definitely a new "feature". The players I looked up earlier with assigned WTN numbers are now in the pro zone.
travlerajm
Talk Tennis Guru
With the update, my doubs WTN and singles WTN are now matching lucky 13.0s.
schmke
Legend
Oh, really? What were their WTNs before?
BallBag
Professional
The men where between 1 and 3. The only ones I wrote down were in the post below.
All those guys are prozone now.
GrassCourtFan
Legend
I looked up somebody, I think Denis Shapovalov maybe, and his WTN was 1 point something, maybe 1.3 or 1.5? Don't remember. I looked up one of the recent French Open Junior men finalists and their WTN was I think three point something, maybe two point something.
schmke
Legend
Actually, I found some. A sampling:
A 5.5C male was a 4.5 WTNs and is now Pro Zone
A 6.0C male was a 3.6 WTNs and is now Pro Zone
A 5.5C male was a 3.5 WTNs and is now Pro Zone
An unrated female was a 10.8 WTNs and is now Pro Zone
A 5.5C female was a 15.1 WTNs and is now Pro Zone
I think they are still trying to figure some things out. Or these players are just getting into a Futures even and getting a ranking point which makes them "Pro Zone" regardless of what their WTN is.
OnTheLine
Hall of Fame
Well, if they are going back that far and they are not weighting much towards newer data .... that is 8 matches vs +/- 300 matches going back 4 years... won't make much of a dent in my dismal record.
FuzzyYellowBalls
Legend
Doh, well, the WTN is only used for going to your local public court or club and finding a match, although I don't think any place is actually using it.
Well I got my first chance to consult WTN ahead of a match today. I was matched up with a fellow 4.0 player I had never played, for a practice singles match. I'm a blue-check 20.2 (18.5-22.0), and I looked him up before the match to find he's a blue-check 16.0 (14.2-17.7).
If we believed those Game Zone ranges we should not even bother to play this match, right? Maybe I'm misinterpreting, but if he's not even close to the lower end of my Game Zone I think WTN is telling me to expect a thrashing.
I showed up anyway and it was good match. I struggled in the heat but managed to win 9 games and lost 6-3, 7-6 in just under 2 hours.
Consulting UTR and even TR (despite the latter averaging in our doubles matches) would have told me that I was an underdog but close enough to have a decent chance. Very disappointing that WTN seems to be relatively useless so far.
If we believed those Game Zone ranges we should not even bother to play this match, right? Maybe I'm misinterpreting, but if he's not even close to the lower end of my Game Zone I think WTN is telling me to expect a thrashing.
I showed up anyway and it was good match. I struggled in the heat but managed to win 9 games and lost 6-3, 7-6 in just under 2 hours.
Consulting UTR and even TR (despite the latter averaging in our doubles matches) would have told me that I was an underdog but close enough to have a decent chance. Very disappointing that WTN seems to be relatively useless so far.
schmke
Legend
Yeah, the stated idea behind the Game zONe is that if you play someone within your GZ, the probability of the match is 35-65% either way. So presumably if you played an 18.5 you'd have a 35% chance of winning and if you played a 22.0, 65%. In this case you played a 16.0 and your GZ's don't even overlap. Now, if the roughly +/-2 GZ is good for +/-15%, one might extrapolate that the 4.2 point difference should be good for about -35%, so you were supposed to have a 15% chance of winning. Given the score and your experience, yeah, that seems a little low.
S&V-not_dead_yet
Talk Tennis Guru
These systems are all about probabilities and confidence bands, not absolutes. Think "opponent has an 80% chance of winning, +/- 3%; OTM's expected game count is 4 games with a standard deviation of 2." If you look at it that way, you're still within range but you scored a number of games which was several standard deviations from the mean. It doesn't mean the system is broken, just that the bands are fairly wide. As the system gets more matches [and they tweak the algorithm], the bands should shrink. They likely run all of the results through a backtester and then make changes to see how that would have affected the ratings and would that have narrowed the bands.
Maybe UTR was like this in the beginning?
If this keeps happening, the gap between his rating and yours should narrow.
schmke
Legend
They've been working on it for at least three years (it was announced in 2019) and have said in various places that 4 years of data is used, or that they used matches back to 2016. So all your points are valid, but shouldn't they have a sufficient number of games and have done the testing you describe on the 4-6 years of data they already have?
S&V-not_dead_yet
Talk Tennis Guru
Fair enough.
Although, it just occurred to me that the high/low of WTN is 40 compared to UTR of roughly 16 [I know there's no theoretical limit]. So WTN's range is 2.5x that of UTR. So the variation we see in WTN has to be divided by 2.5 in order to roughly compare it to UTR. This can make WTN more reasonable [although I've seen plenty of data on these threads where even dividing by 2.5 wouldn't make them more accurate than UTR].
Fully agree, they have enough data (back to 2016) and they don't even include recent data... I have tested it with my entire USTA league team (May and June 2022 matches), the WTN rating is completely useless and incorrect for the moment.
Similar threads
