The ITF has changed the WTN algorithm, is it any better?

schmke

Legend
I got an e-mail from the USTA today, I imagine many folks did, touting the ITF on-boarding more federations, and also updating the WTN algorithm, the implication being it has improved.

My initial look is that it has changed, but I'm not sure it improved. More details on my blog, but personally and for those players I've partnered and played against, it simply appears all the rating got higher (worse rating) but sort of uniformly so. For example my WTNd went from around 22 to around 29 and my partners/opponents went up similarly. A teammate had been in the mid/high teens and is now at 25. And relative ratings between men and women are not noticeably different, meaning it is still broken.

So it appears all they did was a major shift up to make more room for high level players by getting us recreational players out of the teens and into the 20s and 30s.

What are you seeing?
 

Purestriker

Legend
I got an e-mail from the USTA today, I imagine many folks did, touting the ITF on-boarding more federations, and also updating the WTN algorithm, the implication being it has improved.

My initial look is that it has changed, but I'm not sure it improved. More details on my blog, but personally and for those players I've partnered and played against, it simply appears all the rating got higher (worse rating) but sort of uniformly so. For example my WTNd went from around 22 to around 29 and my partners/opponents went up similarly. A teammate had been in the mid/high teens and is now at 25. And relative ratings between men and women are not noticeably different, meaning it is still broken.

So it appears all they did was a major shift up to make more room for high level players by getting us recreational players out of the teens and into the 20s and 30s.

What are you seeing?
That is weird. Mine went 2 points. But you are a 4.5, there is no way you should be near a 3.5.
 

JLyon

Hall of Fame
they did do a major correction, mine went from a 12.3 (way to high) to a 25, which seems more inline, but I think solid 4.5 players are probably should be closer to 17-19 range.
 

cks

Hall of Fame
What are you seeing?
As a 3.5C league player, my WTN was in the mid 20s (i think), and now is reporting in the low 30s. I'm not exactly sure where to view my historical WTN.

Edit: My WTN increased from 24.2 (April 2023) to 31.8.
 
Last edited:

Jack the Hack

Hall of Fame
LOL, mine went from hovering between 17 and 19 for the past several months to 32.3 this week! :-D

I'm not offended. Nobody really knows how this number works right now, so I'm not paying attention to mine or anyone else's until this gets worked out. In the meantime, I'm about to play a string of consecutive singles tournaments for the first time in a few years that should feed the system a bunch of updated data.
 

TennisOTM

Professional
Below are some quotes from the FAQ found here: https://www.usta.com/content/dam/us...Tennis-Number-Algorithm-Changes-July-2023.pdf

"Adult players aged between 19-29 should expect only minor changes to their ITF World Tennis Number. Larger changes begin to occur for players over 30 years old that regularly play against players of a similar age. Adult players over 30 will see their numbers move down the scale (towards 40), this is to create some balance with Junior players and to ensure full use of the 1-40 scale."

"Junior players aged 10 and under will most likely experience a movement down the scale, (towards 40) with players who are aged 17 and 18 moving up the scale (towards 1). The player ratings are now more of an accurate reflection of age, with a general incremental increase expected from aged 10 upwards."


So perhaps the adjustment was mostly an attempt to improve the comparison across different age groups that don't play each other very often? The under-30 vs. over-30 distinction is pretty strange though. Are the 20-something players in USTA leagues just arbitrarily ranked better even if their match results don't reflect that??
 

schmke

Legend
Below are some quotes from the FAQ found here: https://www.usta.com/content/dam/us...Tennis-Number-Algorithm-Changes-July-2023.pdf

"Adult players aged between 19-29 should expect only minor changes to their ITF World Tennis Number. Larger changes begin to occur for players over 30 years old that regularly play against players of a similar age. Adult players over 30 will see their numbers move down the scale (towards 40), this is to create some balance with Junior players and to ensure full use of the 1-40 scale."

"Junior players aged 10 and under will most likely experience a movement down the scale, (towards 40) with players who are aged 17 and 18 moving up the scale (towards 1). The player ratings are now more of an accurate reflection of age, with a general incremental increase expected from aged 10 upwards."


So perhaps the adjustment was mostly an attempt to improve the comparison across different age groups that don't play each other very often? The under-30 vs. over-30 distinction is pretty strange though. Are the 20-something players in USTA leagues just arbitrarily ranked better even if their match results don't reflect that??
Yeah, it appears that way. They've determined there are "islands" of players, or what they call cohorts, that don't have much if any play between them, so no opportunity for the algorithm to self-correct. Presumably they looked at the limited matches between these islands and made the adjustment based on what those results told them.

Here are a few more quotes from the FAQ:

These enhancements to the ITF WTN algorithm will reposition cohorts of players on the current scale of 40-1 and ensure that players from all over the world are more accurately aligned, particularly with regard to different age groups. While ratings are likely to be adjusted, players will likely see movement relative to their age group.

And:

Q: Does this mean my previous ITF World Tennis Number was wrong?
Your ITF World Tennis Number previously was accurate within your network of players and will
remain so with the changes. What will change is your number relative to players outside your
network. The ITF World Tennis Number is designed to accurately reflect the ability of all players, and
these changes will ensure players can always find a competitive match regardless of factors such as
age.
 

Jack the Hack

Hall of Fame
Yeah, it appears that way. They've determined there are "islands" of players, or what they call cohorts, that don't have much if any play between them, so no opportunity for the algorithm to self-correct. Presumably they looked at the limited matches between these islands and made the adjustment based on what those results told them.

This "islands" of age groups thing assumes that people over 30 aren't playing people under 30 very often. Perhaps that is true in other countries, where a 51-year-old person like me might only be playing 50-and-over tournaments and never plays a younger person. However, here in the US, I'm playing 18 and over USTA League as well as NTRP tournaments in addition to age group tourneys. Therefore, I'm encountering plenty of folks under 30.

Anyway, like I said in my previous statement, I am not going to pay much attention to this World Tennis Number. It's meaningless at the moment.
 

Moon Shooter

Hall of Fame

WTN has been a mess but I think this shift is for the better. I gained about 7 points but without this change there was clearly no room for the pros. It is stupid to have a rating system where you can’t include the pros - they still may not include them which is why utr will remain of interest.

But on the whole this is not a large improvement. Like you say ratings between men and women is still broken. This is because usta does not allow matches to correct this like utr does. Even though utr is pretty much 99% based on usta matches in my area the few truly coed doubles they have allowed has done quite a bit to equalize - although that is still off as well. Until we have a substantial number of truly coed doubles matches these rating will remain off. Once that happens the men’s ratings should start improving and the women’s ratings get worse.

and was this really a substantial change to the algo or just a minor tweak that added a certain amount of points based on when you played most of your games. I had noticed that players with earlier games before 2019 or so had incredibly good ratings and you said your singles rating was based on old games. That may be why you got a bigger hit on singles then on doubles.
 

schmke

Legend
But on the whole this is not a large improvement. Like you say ratings between men and women is still broken. This is because usta does not allow matches to correct this like utr does. Even though utr is pretty much 99% based on usta matches in my area the few truly coed doubles they have allowed has done quite a bit to equalize - although that is still off as well. Until we have a substantial number of truly coed doubles matches these rating will remain off. Once that happens the men’s ratings should start improving and the women’s ratings get worse.
The reason for this significant adjustment to some players is they recognized an issue between islands or cohorts of players that were sparsely connected or not connected at all. Like you say, the algorithm can't adjust if there is no data to base an adjustment on, so whether it was purely a manual adjustment, or they found the limited number of matches between the islands and gave them more weight, they made an adjustment.

The men and women are not entirely different, just two different islands of match data, and they could have recognized things were out of whack and done a similar adjustment. But they either A) didn't recognize the issue, or B) don't care.

Or, it could be that they focused on getting the distribution in 1-40 right for men, and are now discovering that there isn't enough of a range for women too. For example, a bunch of 4.5 men found their WTN moved from the teens to the mid/upper 20s. That doesn't leave a lot of room for the 4.0 and below men to fit in the upper 20s and 30s, but lets say they did it and you have 3.0 (and 2.5) men in the high 30s. If you assume the same NTRP male is a bit stronger than the same NTRP female, that leaves no room for the females as they'd need to go into the 40s.
 

ChaelAZ

G.O.A.T.
I was around 21 on both singles and dubs and now 30.

That is a CRAZY shift to try and figure, and doesn't bode well for confidence in a "thought out" system. That kind of correction seems like they had something very wrong. Like Jack said above, not putting my confidence in WTN, it is a mess.
 

Dags

Hall of Fame
So it appears all they did was a major shift up to make more room for high level players by getting us recreational players out of the teens and into the 20s and 30s.
This appears far more inline with the WTN numbers allocated in the UK. I was quite baffled when club players on these boards were posting similar numbers to the performance juniors and senior county players from where I am.
 

Jack the Hack

Hall of Fame
This appears far more inline with the WTN numbers allocated in the UK. I was quite baffled when club players on these boards were posting similar numbers to the performance juniors and senior county players from where I am.

I kind of wondered about the WTN ratings I was seeing for non-US players. For example, Jonas Eriksson, the founder of the TennisNerd website and Youtube channel, is listed as a 29.0 in singles. Yet, when he posts videos of himself playing, he looks more like what I would equate to a 4.5 NTRP in the US, most of which were getting WTNs in the teens. So if 4.5 now equals a WTN range in the mid-20s, I guess that makes some sense. However, that only leaves about 12 points between 40 and 28 for 2.5 to 4.0 players, which is where a bulk of recreational players reside.

Anyway, as I keep saying, this is a total mess until they get more data and figure out the algorithm better.

(In the meantime, I hope opponents look at my meager 32.3 WTN and think I suck before tournament matches. There's nothing like the sinking feeling of underestimating someone, not preparing, and losing to a person you think is inferior. ;) )
 

schmke

Legend
I kind of wondered about the WTN ratings I was seeing for non-US players. For example, Jonas Eriksson, the founder of the TennisNerd website and Youtube channel, is listed as a 29.0 in singles. Yet, when he posts videos of himself playing, he looks more like what I would equate to a 4.5 NTRP in the US, most of which were getting WTNs in the teens. So if 4.5 now equals a WTN range in the mid-20s, I guess that makes some sense. However, that only leaves about 12 points between 40 and 28 for 2.5 to 4.0 players, which is where a bulk of recreational players reside.

Anyway, as I keep saying, this is a total mess until they get more data and figure out the algorithm better.

(In the meantime, I hope opponents look at my meager 32.3 WTN and think I suck before tournament matches. There's nothing like the sinking feeling of underestimating someone, not preparing, and losing to a person you think is inferior. ;) )
Given they are now cramming the majority of recreational players into the 40-25 range, the clear message would seem to be WTN is focusing on juniors, collegiates, and pros giving them the larger 25-1 range to fit in. This isn't necessarily bad, after all NTRP has 5 levels from 2.5 to 4.5, so WTN having around 15 round numbers is finer grained.

But surely the USTA and ITF could have done this analysis to see that recreational USTA League players rated in the teens was not consistent with the UK or didn't allow room for juniors/collegiates/pros and done something about it over a year ago before launch.
 

travlerajm

Talk Tennis Guru
Given they are now cramming the majority of recreational players into the 40-25 range, the clear message would seem to be WTN is focusing on juniors, collegiates, and pros giving them the larger 25-1 range to fit in. This isn't necessarily bad, after all NTRP has 5 levels from 2.5 to 4.5, so WTN having around 15 round numbers is finer grained.

But surely the USTA and ITF could have done this analysis to see that recreational USTA League players rated in the teens was not consistent with the UK or didn't allow room for juniors/collegiates/pros and done something about it over a year ago before launch.
Seems like a botched launch all-around.
 

taylor15

Hall of Fame
Mine went from something like 21.x to 31.2

I looked at a friend who is a 5.0C and UTR ~10. He went from 9.x to 23.4. Must be because he’s 37 years. He’s and ex-collegiate player and now a collegiate coach who has a winning record in Atlanta 5.0 leagues, I was surprised at his shift more than mine.
 

esgee48

G.O.A.T.
These rating groups are dumb. NTRP goes from basically 1.0 to 7.0 max. I don’t really understand why they settled on 0.5 difference for grouping. Why not just make the granularity 0.1 and leave it at that. Then you can declare Open level 5.2 or higher. Levels do not have to be 0.5 apart. Then you can have matches at low, mid or high levels, eg 4.1-4.3, 4.2-4.4, 4.3-4.5. That’s what UTR and WTN have done conceptually.
 

travlerajm

Talk Tennis Guru
It really seems like they could have mapped this out a hell of a lot better from the beginning.

It would also draw a lot more interest if they were transparent about what the ratings mean, and provided a rough guideline “conversion chart” to get buy-in on the new rating system.

For example, they could have said:
1 = Top 10 atp
5 = Top 200 atp
10 = Top 1500 atp
15 = ntrp 5.5 male
20 = ntrp 5.0 male
etc

Instead they have a rating system that isn’t yet ready for prime time.
 
Best doubles player in my area (that isn’t carrying ATP points and regularly plays tournaments/USTA) is a 5.0 former very solid college player, and not quite 30. Best I can tell has been in the low 20s for at least the last year.

I’m a 3.5s, also not quite 30, I was as low as 28 and am now 29.9. Interesting the swing difference between areas.

Any ideas why that would be?
 

travlerajm

Talk Tennis Guru
Best doubles player in my area (that isn’t carrying ATP points and regularly plays tournaments/USTA) is a 5.0 former very solid college player, and not quite 30. Best I can tell has been in the low 20s for at least the last year.

I’m a 3.5s, also not quite 30, I was as low as 28 and am now 29.9. Interesting the swing difference between areas.

Any ideas why that would be?
I’m trying to make sense of the new WTN, but unfortunately I can’t find much consistent logic in it yet.
 
I’ve been pondering this the last few hours and am currently watching Winston Du handle a female four star recruit. I still think leaving over half the ranking for high level juniors and college players isn’t great. I’ve seen way too many good 4.0s and 4.5s beat good junior players. Obviously the best of the best junior players are going to play top three or four for an ACC school are going to be a cut above 5.0s, but it seems the new scattering of ratings is almost intentionally creating more “islands” for players that will lead to an empty 20-25 (for example) and then a big jump to elite juniors and mid-tier college players, then possibly another gap before high level collegians and low level pros, and then another gap to high level pros. Seems like they’ve moved the issue not addressed.
 

ChaelAZ

G.O.A.T.
Just a quick search of teammates and opponents of last season's 4.0 league, looks like the top guys that were borderline high 4.0 to low 4.5 end up around 24-25 WTN, and lower 4.0 to high 3.5 are around 30 WTN, so maybe 4.0 range is like 29-25ish.

Anyone else checking?
 

Dags

Hall of Fame
I had an email from the LTA today, stating that there have been changes to the way WTN is calculated. This was the link they provided for anyone interested (it offers little in the way of explanation):


I guess the change was global. I don't recall what my number used to be, so can't confirm whether it moved up or down! Truth is, adult recreational players have pretty much zero use for ratings over here. I imagine 99% neither know nor care about it.
 

BallBag

Professional
The system that WTN is based on has probabilistic underpinnings which means that the level distribution across the range is not arbitrary. I don't know the actual numbers but a given difference in level should have the same predicted outcome anywhere in the rating range. So if a WTN 5 beats a WTN 7 80% of the time, so should a 13 beat a 15 80% of the time and 33 beats a 35 80% of the time. They are able to shift a whole "rating island" because everyone on the island moves the same amount. They shouldn't be able to squeeze the USTA island from a range of 20 levels to a range of 10 levels.
 

schmke

Legend
The system that WTN is based on has probabilistic underpinnings which means that the level distribution across the range is not arbitrary. I don't know the actual numbers but a given difference in level should have the same predicted outcome anywhere in the rating range. So if a WTN 5 beats a WTN 7 80% of the time, so should a 13 beat a 15 80% of the time and 33 beats a 35 80% of the time. They are able to shift a whole "rating island" because everyone on the island moves the same amount. They shouldn't be able to squeeze the USTA island from a range of 20 levels to a range of 10 levels.
Well, I think there was both a shifting, and an algorithm change as not all players were shifted the exact same amount.

And they did have to do some squeezing of some sort as there were league players in the low teens, even some single digits, to above 30 before the shift, so say 8-35. If everyone was shifted up 10 (roughly the average shift it appears) then range would be 18-45 which goes beyond the cap of 40, so yeah, some squeezing was done.
 

BallBag

Professional
Well, I think there was both a shifting, and an algorithm change as not all players were shifted the exact same amount.

And they did have to do some squeezing of some sort as there were league players in the low teens, even some single digits, to above 30 before the shift, so say 8-35. If everyone was shifted up 10 (roughly the average shift it appears) then range would be 18-45 which goes beyond the cap of 40, so yeah, some squeezing was done.
I guess I should have said that they shouldn't be able to squeeze the ratings while maintaining the integrity of the data. Given all the issues that WTN is having I am much more impressed with UTR, although I wasn't really paying attention when UTR started out. Maybe they were having similar issues.
 

travlerajm

Talk Tennis Guru
I guess I should have said that they shouldn't be able to squeeze the ratings while maintaining the integrity of the data. Given all the issues that WTN is having I am much more impressed with UTR, although I wasn't really paying attention when UTR started out. Maybe they were having similar issues.
Given all the issues with WTN and UTR, I am much more impressed now with TR.
 

Tweener

Semi-Pro
The new WTN ratings are horrible. I've recently started playing mixed doubles matches for the first time and have gone 5-0, primarily due to my play. My partner was decent, but she struggles with returning big serves and is highly inconsistent. She roughly has a .500 record overall but many more matches on her record. Somehow she's a 27.xx and, with the recent changes, I was bumped down to a 34.1. Is my low doubles rating only due to only having 5 doubles matches played? We've beaten teams where both players were in the 27-28 range with scores like 6-4 6-3. On UTR, my doubles rating is double that of my partner's!
 
The new WTN ratings are horrible. I've recently started playing mixed doubles matches for the first time and have gone 5-0, primarily due to my play. My partner was decent, but she struggles with returning big serves and is highly inconsistent. She roughly has a .500 record overall but many more matches on her record. Somehow she's a 27.xx and, with the recent changes, I was bumped down to a 34.1. Is my low doubles rating only due to only having 5 doubles matches played? We've beaten teams where both players were in the 27-28 range with scores like 6-4 6-3. On UTR, my doubles rating is double that of my partner's!
Mixed matches seem to be a real conundrum for WTN, especially with a mix of people who only play mixed and other who play men’s/women’s
 

toby55555

Hall of Fame
This "islands" of age groups thing assumes that people over 30 aren't playing people under 30 very often. Perhaps that is true in other countries, where a 51-year-old person like me might only be playing 50-and-over tournaments and never plays a younger person. However, here in the US, I'm playing 18 and over USTA League as well as NTRP tournaments in addition to age group tourneys. Therefore, I'm encountering plenty of folks under 30.

Anyway, like I said in my previous statement, I am not going to pay much attention to this World Tennis Number. It's meaningless at the moment.
In UK most results probably come now from LTA run Local Leagues which have no age categories. My league is 12 divisions of 8 players played in courts within a roughly 3 mile radius. My opponents in Div. 1 are 15-40 years younger.
 

Tweener

Semi-Pro
Apparently the new WTN ratings also struggle with new players. I played a guy who was playing his first tournament, and he won his first match 6-3 6-0. I beat him 6-0 6-1. After the post tournament ratings were updated, he's at a WTN 25.1 and I'm at a 27.8. In 19 singles matches this year, I've dished out 6 bagels and 5 breadsticks and I'm 14-5. How he's at a 25.1 is beyond me.
 

Tweener

Semi-Pro
WTN doest count games. So all they count is win in two and loss in 2. With only two matches, all WTN can say with reasonable accuracy is that the guy probably plays tennis.
Seems strange that this newly improved system only looks at matches/sets and not games or scores. This affirms my stance that it's pretty useless. I played a 29.6 guy who was way better than this guy who has a 25.1. Similarly with the doubles situation, as a 34.1 (since I only have 5 doubles matches, all straight sets wins), I've played teams with both players in the 27s who were pretty weak.
 

travlerajm

Talk Tennis Guru
The description on WTN site about using info at set level is unclear. I took it to mean they do use the score of each set.
 

loveallcats

Rookie
I wonder why they wouldn't count games. It seems so much more accurate to count games not just sets. What if the set is really close... That's a big difference from a 6-0 blowout that is not being accurately reflected.
 

travlerajm

Talk Tennis Guru
I wonder why they wouldn't count games. It seems so much more accurate to count games not just sets. What if the set is really close... That's a big difference from a 6-0 blowout that is not being accurately reflected.
I interpret info “at set level” to mean they are counting games.
 

BallBag

Professional
WorldTennisNumber said:
WTN also analyses match at set level, meaning the algorithm takes into account each individual set as its own result. Simply, if a match ends 2 sets to 1 in your favour, then the system will update your Number with two set ‘wins’ and one set ‘loss’. Even if you don’t win the overall match, the sets you have won will be considered accordingly in the WTN calculation.
 

Moon Shooter

Hall of Fame
The new WTN ratings are horrible. I've recently started playing mixed doubles matches for the first time and have gone 5-0, primarily due to my play. My partner was decent, but she struggles with returning big serves and is highly inconsistent. She roughly has a .500 record overall but many more matches on her record. Somehow she's a 27.xx and, with the recent changes, I was bumped down to a 34.1. Is my low doubles rating only due to only having 5 doubles matches played? We've beaten teams where both players were in the 27-28 range with scores like 6-4 6-3. On UTR, my doubles rating is double that of my partner's!

It appears at least in the USA the usta or wtn gave all the men much worse WTNs ratings enmasse but not the women. So they made a bad situation even worse. Maybe usta is trying to peg the WTN to their ntrp and are pretending a 4.0m is the same as a 4.0 f. But they even went beyond that so now 3.5 women have the same WTN range as 4.0 men.
 

TennisBro

Hall of Fame
What are you seeing?
My observation now is that some powers in tennis have taken control over the ratings as "their" tournaments count and "their" players benefit from playing those sanctioned matches that are eligible/entitled to be counted in the ratings. I'd go as far as to say that the ratings are being monopolized purposely. Some tennis teams want recognition; clubs/coaches want their players to be sponsored/recongized and some sponsors want to have "a prove of life" they ar justly in the business there too. Ratings aside ITF rankings are becoming an image, a business circus. The WTN and UTR appear like an international group consensual xxx.

Why would some have such an interest to influence the tennis ratings? A simple answer is the sponsorships. Brand names are in the game and so are universities recuiters. Junior competitive players/junior ITF players are dependant on their rankings and the UTR rating assists them into universities and with sponsorships too. When rankings don't work out, the tennis rating may. There are a lot of players out of this loop and may just experience the collateral damage in the process.
 
Last edited:
Top