Interesting, thanks! Maybe the WTN "prediction" was not quite as horrible as I thought. A 15% chance of winning is worth showing up for. I'm curious what would be the equivalent 35-65% match win probability range for dNTRP? My opponent is 0.24 higher than me according to TR - would he be in my Game zONe if they had the equivalent version of that?
Share your WTN - Crowd sourcing the NTRP to WTN mapping
- Thread starter schmke
- Start date
Interesting, thanks! Maybe the WTN "prediction" was not quite as horrible as I thought. A 15% chance of winning is worth showing up for. I'm curious what would be the equivalent 35-65% match win probability range for dNTRP? My opponent is 0.24 higher than me according to TR - would he be in my Game zONe if they had the equivalent version of that?
schmke
Legend
If you just scale things for the ranges, with NTRP being 2.5 to 7.0, 0.24 is 5.3% of that range. WTN is 40 to 1 so the 4.2 difference is over 10%. So WTN definitely has you farther apart relatively.
Moon Shooter
Hall of Fame
He may have an issue with his name being the same as someone else's in the world or something like that. Hopefully WTN will post the matches they are including as well as the performance rating they assign for that match - like TR does. That will tell us more. What is odd is if he was underperforming for his current rating in 2022 then he must have had an even stronger rating at the start of 2022.
I would love a rating system that was upfront enough to simply say if you are 1 point higher you should win the set by 6-2. We know WTN, NTRP and UTR will never be able to do that do that because they have a ceiling and a floor.
I also know a 3.5 that has 3.8 WTN in doubles. It is not blue checked though. However he has played quite a few matches so I suspect they know he has some sort of issue. His singles was blue checked last week now it is not blue checked at 16.3.
As to whether the clear mismatch between men and women can effect doubles depends on the algorithm. If they tend to give more credit to the stronger player for a big win than the weaker player, that can play a role. Also the ratings will be off early on depending on who played mixed or same gender matches to establish a rating. These inaccuracies can then be perpetuated and even exaggerated. It will also depend on whether the mismatch between men and women remains consistent throughout the rating spectrum. For example is a 17 doubles female really a 20 doubles male (3 points too strong) but a 24 doubles female is actually a 29 doubles male (5 points too strong)?
Once they start getting data that includes coed events (mm v. mf or mm v ff etc) this should work itself out relatively quickly. I think men players will get some easy rating points by playing women in singles and doubles. But then when they start playing other men as well they will in effect be giving those free rating points to the men they play.
I hate that they have a low rating be stronger! I guess I should have said men players will easily shed some rating points instead of easilly get rating points. And is talking about a "higher rated" player now supposed to mean a "weaker rated" player? Why make this purposefully obtuse?
Yeah, I don't get why all these rating systems don't use the most natural metric, which is percentile rank.
For example, a rating of 50 means you are better than 50% of players, a 75 means you are better than 75%.
You would need to use maybe 3 decimal places for it to be meaningful at the upper end, since all D1 and pros players would be 99+. But so what. And for rec players, most of us are going to be in the 25 - 95 range, so that's plenty of granularity. I'm assuming the bottom 25% would be casual players who rarely play organized tennis.
S&V-not_dead_yet
Talk Tennis Guru
Paradoxically, for a system to simply do this would require a more complex algorithm.
No system does that, not even your oft-referenced Elo. When two players meet, there is an x% chance playerA will win, a y% chance playerB will win, and a z% chance of a draw.
Tennis gets more complicated because of games. A chess win is the same whether I played a brilliant game, my opponent blundered, he lost on time when I was busted, etc. But in order for you to get your "you should win the set 6-2", the system needs to now take into account relative strength to predict a match score, which would make things exponentially more difficult.
Moon Shooter
Hall of Fame
Do you have an idea of how the percentages vary by game set and matches? If I win 70% of the games against someone then I would think I will likely win over 80% of the sets and over 90% of the matches. So it would seem predicting a match out come should be easier than predicting a set and predicting a set should be easier than predicting a game.
Is WTN predicting these probabilities for the chances of you winning a match or a set?
USTA says their algorithm is a bit over 76% correct at predicting *match* wins.
Honestly I am not sure that is anything to brag about. You have a certain number of matches that are rating blow outs (say .35 or more difference between singles or .5 or more difference on the sum of doubles) so that should be pretty easy to hit above 95%. Even though a small percentage of matches will only involve a tiny difference you should still not be below a 50%.
loveallcats
Rookie
I'd love to know this too, and if they put more weight on 3 setters vs 8 game pro-sets for example.
Moon Shooter
Hall of Fame
tennis and chess do this already. They both predict a certain score and then adjust ratings according to the results. That is how the Elo system works. The only difference is chess can have a set rating point difference equating to a certain score ( for example if you’re 200 points higher it can means you should win 3/4 of the games). NTRP utr and wtn can’t because they have ceilings and floors. So even if a 1.02 gets blown out by a 1.01 he can’t drop below 1.00 so the rating can’t keep a consistent distance throughout. The same happens at the top of the scale for these tennis ratings.
BallBag
Professional
They say 4 years but they don't list the match history yet. We don't know if they differentiate between a full third set, a third set TB or an 8 game pro-set.
The algorithm doesn't support an artificial cap at 1. It will happily go past 1 and zero and into negative numbers. I think the assumption is that no one will go lower than 1 and that's probably how they set their initial conditions for the pros. They are hiding pro numbers now so we might never know.
Moon Shooter
Hall of Fame
"ITF World Tennis Number is simply a scale from 40-1 (we couldn’t resist including our favourite tennis number!). As a complete beginner you’ll receive a Number of 40 which you can gradually start to improve as you play. On the other hand, if you’ve been perfecting your backhand for years, you’ll be closer to 1."
World Tennis Number - Powered by ITF
worldtennisnumber.com
So you are saying the scale actually goes from 40 into negative numbers? I hope that is the case and ITF faqs are so often poorly worded that this seems possible. But I do have to ask what makes you think the ITF goes into negative numbers? Don't you agree the quoted language seems to suggest that the best you can do is approach 1 as a limit?
schmke
Legend
I think WTN has a problem with the floor of one, there are some "recreational" players in the 1's and 2's which obviously leaves no room for actual to pros. Their solution is to simply call anyone with points on the ATP/WTA tours a pro and not show their WTN and instead list "Pro Zone"."ITF World Tennis Number is simply a scale from 40-1 (we couldn’t resist including our favourite tennis number!). As a complete beginner you’ll receive a Number of 40 which you can gradually start to improve as you play. On the other hand, if you’ve been perfecting your backhand for years, you’ll be closer to 1."
World Tennis Number - Powered by ITF
So you are saying the scale actually goes from 40 into negative numbers? I hope that is the case and ITF faqs are so often poorly worded that this seems possible. But I do have to ask what makes you think the ITF goes into negative numbers? Don't you agree the quoted language seems to suggest that the best you can do is approach 1 as a limit?
This means that one of two things is happening. Either they aren't even calculating WTN for pros, or if they are they aren't enforcing the floor and letting ratings go below 1.0 and using Pro Zone to hide this fact. I don't think they can be calculating ratings for pros and enforcing the floor as the 5.0s and 5.5s I see in the 2's, and the 6.0s and 6.5s in the 1's should simply not happen in that case.
BallBag
Professional
Glicko-2 has an offset and multiplier to convert the rating into the Glicko-2 scale. Glicko-2 scale is 0 for the average value around 10 for 3 sigma values. For chess its:"ITF World Tennis Number is simply a scale from 40-1 (we couldn’t resist including our favourite tennis number!). As a complete beginner you’ll receive a Number of 40 which you can gradually start to improve as you play. On the other hand, if you’ve been perfecting your backhand for years, you’ll be closer to 1."
World Tennis Number - Powered by ITF
So you are saying the scale actually goes from 40 into negative numbers? I hope that is the case and ITF faqs are so often poorly worded that this seems possible. But I do have to ask what makes you think the ITF goes into negative numbers? Don't you agree the quoted language seems to suggest that the best you can do is approach 1 as a limit?
µ = (r − 1500)/173.7178
They didn't publish it for WTN but I thinks its:
µ = -(r − 20)/2
So if you are 3100 on chess.com then its 9.21 on the Glicko-2 scale and if you are a 1 WTN then its 9.5 Glicko-2 scale. There is nothing in the algorithm stopping from going higher than 3100 on chess.com but you do have to beat a bunch of 3100 rated players and that might be hard.
BallBag
Professional
I'm guessing they initialized the pros so that ATP #1 is around WTN 1 but those pesky league players are not playing on tour so they started creeping into pro territory.
schmke
Legend
This shows the challenge of pockets or islands of unconnected players/matches and yeah, whatever initial mapping or guess they used along with how the algorithm is set up allowed ratings "encroachment" that isn't representative of actual levels. But they had all this info before things were published so you would have thought they'd adjust things. But my guess is the ITF had WTN "done" already (LTA started using it last year I think) and the USTA was late getting their matches in and discovering the problem and it was too late to adjust things in the algorithm, so they've just gone with it and don't tell us the pro's WTNs and have crossed their fingers that non-pros won't get all the way to 1.0 ...
BallBag
Professional
I think the problem is that they are seeing wider than expected distribution of USTA ratings. They can either push all US players closer to 20 or shift US player down a few levels.
Moon Shooter
Hall of Fame
Thanks I only have a rudimentary level of understanding of these different algorithms but I think WTN nor UTR modify the Glicko-2 system. And since there are an infinite number of real numbers between even 0 and 1 (it doesn't matter whether the scale goes from 0 to 1 or from 0 to 10,000 the issue is the cap and floor) it would seem they could incorporate a floor and ceiling in these rating systems but they would have to have some sort of diminishing returns as someone got closer to the ceiling (or floor) and that would effect ratings everywhere in between. I think UTR does this and that is, in part, why I think UTR ratings will always jump around so much. Everyone else's games seem to reshuffle your own placement in this grid from 1-16.5. Notice that the top players are all approaching 16.5 but never quite get there even if they have a great run. I think they are putting a limit at the top.
Now Tsitsipas at 15.89 is .15 UTR points higher than Korda 15.75. I think at the adult rec level being .14 different -even if we played just as many matches as the pros (in the past 12 months) would mean a closer game would be predicted. But at that upper extreme I think that they started to hit a limit at 16.5 so the numerical differences between the very top players is less than the actual skill level. I am not saying there is a huge difference between Korda and Tsitsipas but I think I would pretty clearly bet one Tsitipas with even money. But if you told me someone rated 6.89 was playing someone 6.75 (again even if they played the same number of matches in the past 12 months as Korda and Tsitsipas) I doubt we would see much difference. Of course there is some consideration of the different surfaces they just finished the clay season and I would think Tsitsipas would have an even bigger advantage on clay.
Consider Bobby Fischer at the time had the highest rating of 2785 and Carlsen had a rating of 2864. Even if the rating system did not change, that does not necessarily mean he played the strongest chess of all time although that is likely true. But the FIDE rating never even tried to put a ceiling on its rating. So whether you are 2219 playing a 2019 or a 1219 playing 1019 the difference of 200 points is the same so the expected score is the same. (for example the higher player may be expected to win 3 out of 4 games in both cases)
But it seems the tennis systems all like to have a ceiling and floor. And it seems a system that has a floor and a ceiling is unable to
1) account for the possibility that someone is playing (insert game here) better than anyone in history
And at the same time
2) keep a set standard where a certain rating difference equates to a set predicted result. For example if someone is 2 rating points weaker they would be expected to lose a set 3-6.
Moon Shooter
Hall of Fame
Just to illustrate the problem of pockets. My singles rating on utr after 5 matches is currently 2.14. To be fair they say it is only 40% reliable. I am 2-3 and all of the matches were against 3.5 level players except one guy that is not rated - I posted a few of my matches against him from prior years in the post your video thread and we have both improved since then. But when I look at some of the 3.0 players - one of which I just beat in a practice match 6-0 4-0 (we stopped playing to switch it up - and to be fair I don't think I won any game by more than the minimum score of 2 points) they are all in the 4.5-6.5 range! In other words these hard and fast USTA rules that divide who can play each other creates pockets.
I really think this can be the beauty of this WTN if they allow us to truly have a variety of players in these tournaments and also not cut off the matches that count at 12 months. I think that 12 month cut off simply dooms UTR from ever being a decent rating system for adult rec players.
I really think this can be the beauty of this WTN if they allow us to truly have a variety of players in these tournaments and also not cut off the matches that count at 12 months. I think that 12 month cut off simply dooms UTR from ever being a decent rating system for adult rec players.
BallBag
Professional
UTR is not based on Glicko-2. They have their own proprietary algorithm that normalizes the range between 1 and 16.5. For UTR, the expected score for a 0.5 difference is 6-3 6-3. They might have updated the algorithm since that was shared with me.
UTR counts the .5 combo matches that NTRP does not so there is some mixing there. Also there are people playing up and people moving up and down the NTRP ladder. Adult players are hard to rate for any system because they don't have much data.
Last edited:
Moon Shooter
Hall of Fame
They can't do both. They cant fit everyone in between 1 and 16.5 and also make it so a 1.5 difference equates to a 6-3 6-3 score. You can only do one or the other.
The lack of data is a problem. But I think the bigger problem is the amount of data that NTRP ignores. NTRP ignores all mixed matches (not just the combo ones that end in .5) if you have a adult rating for the same gender. It will also ignore data that is not from the rigidly structured USTA matches. WTN will not only create more data because it has much less rigid requirements for setting up matches, it also doesn't ignore the data from the matches that do happen. WTN may have some issues at the start, but it will soon be a much more accurate rating system than NTRP.
The only way NTRP sort of includes more data is because it doesn't completely separate singles from doubles. But if WTN is clear about the number of matches that are going into someone rating I think people can get a better sense of a players ability by looking at the doubles and singles than NTRP alone would provide.
BallBag
Professional
That was a typo. It's 0.5 for 6-3 6-3. I don't know how WTN is going to get more data than it already has. As new match data comes in, old data will get bumped out.
Moon Shooter
Hall of Fame
That sounds closer but they still can't do that if they keep a ceiling and floor on the ratings. You can either peg the ratings so everyone stays between two numbers, or you can peg the ratings so a certain difference in rating is set to a predicted score. But you can't do both.
They will allow coed matches (not just mixed MF v MF) and a larger variety of ages to play together. They also may allow non-usta members to play in events that will effect your ITF rating. Plus players that just finished college or high school will not be prevented due to artificially high self rate minimums. With a larger pool of players and more flexibility it should be easier to get competitive events organized. It is still unclear how much flexibility they will give organizers it is still pretty clear there will be more than the current USTA structures allow. More events equals more data. NTRP will likely still ignore these new events and so their ratings will continue to be less informative/accurate.
BallBag
Professional
That was the premise for UTR and I'm not seeing a lot of good coed events in my area. Maybe WTN will have better luck.
Moon Shooter
Hall of Fame
I am running an informal singles group where we plug in our scores into UTR. 2 of my 5 singles matches are from that league. (I have to finish up one and schedule 3 more that were postponed - and I am sort of swamped with USTA commitments now) But only the players can enter their scores and many don't want to be bothered so it doesn't often get done. Plus it is not really a full tournament structure with I think some players like.
In any case I will probably play 2 more singles matches from USTA this year. If it weren't for WTN I think I would have gotten several times as many in UTR - about 5-10 per season. The problem is we are just playing each other in UTR and don't have much of an anchor to non-local players. And the few anchors we do have seem way off. Now with WTN seeming to have a much more sensible rating, UTR really does seem like a waste of time. UTR is always changing (is it because everyone must always be rated between 1 and 16.5?) and then every match is completely erased after 12 months! They just made some foolish decisions when they designed the rating and seem to be doubling down on the bad ideas.
I am cautiously optimistic that WTN will be a big reason for people to participate in competitive adult rec tennis.
FuzzyYellowBalls
Legend
I don't think WTN has the ambitions this thread is giving it, it certainly won't push NTRP out of USTA events and it won't create its own events ala UTR, so it is basically an international cross-gender number to use if you want to try and find a one off match while travelling or at your local park or club.
schmke
Legend
I'm pretty confident the USTA has designs on having tournaments, certainly at the junior level, using WTN just how UTR is used for its events. You are right it won't push NTRP out of Leagues, but I would not be surprised to see them pushing WTN for adult tournaments.
FuzzyYellowBalls
Legend
Oh yeah, the youth market, not being a coach I always forget that, decent money there too, that's how USTA is eyeballing the market UTR and how to capture it, makes sense.
Moon Shooter
Hall of Fame
I don't think anyone is saying it will push NTRP out of same gender league play. But, assuming they avoid any foolish moves, WTN will replace NTRP for tournaments. NTRP has completely different scales for men and women so unless you want a tournament that is all men or all women NTRP is practically useless. Can you play the opposite gender in USTA tournament matches now? If so will it effect your NTRP?
Will USTA ever offer a coed league? If so WTN would seem the only option.
Since WTN includes more data than NTRP it will almost certainly be more accurate than NTRP.
The UTR rating system has too short of an attention span, and lacks the transparency, for people to really be motivated at improving their UTR. Unless you and those around you play a ton of rated matches every year UTR is like your random number of the day.
FuzzyYellowBalls
Legend
Just for your collection of rec league tennis information. Some markets might be shut off from a USTA co-ed league. Dallas has the female focused TCD (Tennis Competitors of Dallas) which runs mostly ladies leagues that are more popular by far than USTA and a mixed section. Has to be going back decades, so a USTA one won't gain much traction. I imagine Atlanta also has something similar. Perhaps others can chime in about other cities. Not a big deal, just information.
Moon Shooter
Hall of Fame
I think having gender separated leagues and tournaments is great, if your community has enough players to support them. I don’t think wtn will have much effect on the Atlanta or Dallas tennis scene for several years. I think the wtn can help smaller tennis communities that need to scrounge all the players they can for their events.
a good rating system integrated into easy to use software that allows for reasonably priced , flexible, and usta endorsed events I think sounds promising.
It may not work for the reasons similar to those cashman offers. But I think this is a very good attempt. It’s hard to knock what what usta is trying to do.
ChaelAZ
G.O.A.T.
They could use NTRP, but they would need to fully integrate co-ed to level everyone out on the same scale - meaning you could not have any men v men only and women v women only leagues or you would still have the differentiation. So in that regard, WTN is a much better solution allowing everyone to still compete in sex'd leagues or tourneys with NTRP, and then do coed tourneys or leagues even using WTN.
Personally, that is one thing I liked about UTR, because I am just looking for level competition, be it some guy I know or a 16 year old female teenie bopper being all woke and texting her friends after about the old dude she just had to play. As long as the points are good, I am in.
Moon Shooter
Hall of Fame
No changes to my rating. It would be nice if they let you know what matches are included.
San Diego county also has a lot of non-USTA leagues for women that are very popular. I would say half the women play USTA leagues too and half play exclusively local leagues.
FuzzyYellowBalls
Legend
Women are smarter than us, they are like f- usta, we'll make and run our own leagues with our own rules. Little tennis kingdoms.
PurcellPower
Rookie
3.0C
WTNs - 18.8, GZ 20.6-16.9
WTNd - 21.2, no GZ
WTNs - 18.8, GZ 20.6-16.9
WTNd - 21.2, no GZ
ChaelAZ
G.O.A.T.
Pretty stable as expected. UTR ticked a little, but no effect on WTN or NTRP, of course.
UTR 5.75 sgl / 5.3 dubs
WTN 20 gl / 21.6 dubs
travlerajm
Talk Tennis Guru
Something that would be interesting and useful:
I’d be curious to know what WTN values correspond with the boundaries between ntrp levels.
For example, the boundary between 4.0 and 4.5 level might be defined objectively as the WTN value that gives the minimum number of combined overlap players on both sides of the boundary.
If the boundary between 4.5 and 4.0 is, say, 18 WTN, then there will be x number of 4.0 players with WTN < 18, and y number of 4.5 players with WTN > than 18. 18 is the value that minimizes the sum of x + y.
Edit: thinking about this more, the minimum should be done the way thresholding is done for image processing. That is, the valley between the category-population-normalized histograms is found by minimizing overlap area under the curve.
I’d be curious to know what WTN values correspond with the boundaries between ntrp levels.
For example, the boundary between 4.0 and 4.5 level might be defined objectively as the WTN value that gives the minimum number of combined overlap players on both sides of the boundary.
If the boundary between 4.5 and 4.0 is, say, 18 WTN, then there will be x number of 4.0 players with WTN < 18, and y number of 4.5 players with WTN > than 18. 18 is the value that minimizes the sum of x + y.
Edit: thinking about this more, the minimum should be done the way thresholding is done for image processing. That is, the valley between the category-population-normalized histograms is found by minimizing overlap area under the curve.
Last edited:
schmke
Legend
You've seen the charts I posted? There is a lot of overlap, the "minimum" overlap won't be very small at all.
travlerajm
Talk Tennis Guru
The charts indeed havemuch overlap, and that’s why I thought it would be interesting to estimate where the boundaries roughly are. Otherwise, I feel lost with my WTN, because I don’t know what it really means.
travlerajm
Talk Tennis Guru
The chart tells me fairly definitely that I’m above average for a 4.5, and below average for a 5.0.
The missing piece of information that I think many others would be interested in for their own situation:
How close be being over the hump, on the 5.0 side of the 4.5-5.0 boundary, am I?
travlerajm
Talk Tennis Guru
I’m spoiled and appreciative of @schmke’s number crunching and analysis.
Assuming we have an appropriate function for determining the degree of overlap, then a cool way to present the data would be plotting the degree of overlap as a function of WTN.
This would give you a chart with a series of hills, with the peak of each hill representing the avg of the given ntrp level, and the valleys between hills defining the boundaries between ntrp levels.
It’s clear to me whether or not the hills will line up with the hills on the original chart?
Assuming we have an appropriate function for determining the degree of overlap, then a cool way to present the data would be plotting the degree of overlap as a function of WTN.
This would give you a chart with a series of hills, with the peak of each hill representing the avg of the given ntrp level, and the valleys between hills defining the boundaries between ntrp levels.
It’s clear to me whether or not the hills will line up with the hills on the original chart?
schmke
Legend
I'm not sure the approach you outline really works when there aren't a similar number of players at each level.
There are a lot more 4.0 men than 4.5 men, so to minimize the x+y you mention requires moving the WTN to such a high number that all the 4.5s are below it and there are no 4.0s above it.
For example, with the data I have from the first week, a WTNs of 20.0 has 8974 4.0 men higher and 15,059 4.5 men lower for a total of 24,033, but a WTNs of 21.0 has 6,551 4.0 men higher and 15,644 4.5 men lower for a total of 22,195. That x+y is significantly lower but it keeps going lower the higher I go since going higher removes 4.0s faster than it adds 4.5s, and I don't think that would suggest the 4.0/4.5 boundary is 25.0 or something.
What perhaps makes more sense is to just take the average and median WTNs for each NTRP level and pick the midpoint between adjacent levels as an estimate of the boundary.
3.0 - avg 24.0, median 23.8
3.5 - avg 21.2, median 20.7
4.0 - avg 18.1, median 17.5
4.5 - avg 15.0, median 14.1
5.0 - avg 12.1, median 10.9
5.5 - avg 12.0, median 9.9
We see the difference is ~3 until you get to 5.0/5.5, and taking the midpoints you get:
3.0/3.5 - 22.6 or 22.2
3.5/4.0 - 19.6 or 19.1
4.0/4.5 - 16.5 or 15.8
4.5/5.0 - 13.5 or 12.5
schmke
Legend
@travlerajm Of course, if we look at how many players by level are on either side of these ...
Here are the thresholds based on average:
22.6, the 3.0/3.5 boundary - 64% of 3.0s are higher and 36% are lower. But there are a full 31% of 3.5s that are higher, and even 13% of 4.0s that are higher than the 3.0/3.5 threshold. And even 6% of 4.5s are higher! Being higher means these 4.0s and 4.5s are on the 3.0 side of the 3.0/3.5 boundary!
19.6, the 3.5/4.0 boundary - 6% of 3.0s are lower than this which is reasonably few. But 31% of 4.0s are higher, and 13% of 4.5s are higher, and even 9% of 5.0s are higher and on the 3.5 side of the boundary.
16.5, the 4.0/4.5 boundary - 5% of 3.5s are lower than this which is reasonably few. But 28% of 4.5s are higher and 16% of 5.0s are higher and on the 4.0 side of the boundary. And going the other way, 5% of 3.5s are under which has them on the 4.5 side of the boundary. And 0.4% of 3.0s are which is very few, but not zero.
13.5, the 4.5/5.0 boundary - Here we have 6% of 4.0s are lower and 43% of 4.5s are lower.
socallefty
G.O.A.T.
It is the same in Orange County, but the local leagues (PacSun, Hill and Canyon) are weekday morning leagues and it is mostly housewives and retired women who play in it. The women with weekday jobs seem to play in weekend USTA leagues including mixed doubles. Evening WTT team tennis leagues on weekdays seem popular with younger women without kids.
The women playing weekday leagues socialize a lot together after matches and with team parties in addition to doing group team practice sessions together multiple times a week. So, it seems to be a huge part of their social lives throughout the week.
travlerajm
Talk Tennis Guru
Thanks. This helps clarify the picture.
One last question: what % of 5.0s are above the 13.5 boundary?
BeyondTheTape
Semi-Pro
im very curious about what matches have been calculated. how am i a 7.5 !?! 
schmke
Legend
29% of 5.0s over 13.5, 71% under.
ChaelAZ
G.O.A.T.
I saw this happen with UTR when it first came out. There seemed to be a huge drop for players, one quick drop, then like 1.5 to 2 points total over a few months. Maybe see the same in WTN.
travlerajm
Talk Tennis Guru
Yeah. I sense that WTN is still working the kinks out of the algorithm and it hasn’t settled yet.
Similar threads
