Understanding the Tennis record algo

[southernncgal]

For singles, your match rating is based on your opponent's pre-match rating and the score of the match only (your own pre-match rating doesn't matter). For the score it should only count the number of games, so your 6-3, 0-6, 0-1 result would be counted as 6-10 in games. If the game score counted that way had been tied then your match rating would be the same as your opponent's pre-match rating (3.19). But you lost more games, so your match score would be something less than 3.19. To get the "something" you can browse around TR to find a rated match where the game score was 6-10, and figure out the answer that way.

Thanks, from others and other items I've read, I think both players pre-match ratings count so that USTA gives an "expected result" of the match. So, if I'm a 2.75 and she is a 3.19, the expected result would be 6-0 6-0 given we're so far apart to begin with. Getting 6 games to her 10 would beat the expected result, but then I'm lost as to how the match is calculated.

 

[southernncgal]

I'm not Shmeke or anywhere close to his awesomeness, but I've managed to pretty closely reproduce the tennis record results with my own Excel sheet, using the info about partner's and opponent's ratings as a starting point (so obv if that is wrong, my result will be wrong too). I've compared my results to tennis records (I run mine match by match before theirs come out) and it's pretty darn close.
I suspect you can create a similar calculation in Excel if you follow some of the assumptions that are out there:
*Between 3.0 and 3.5 expected score is 6-0, 6-0 therefore each game is worth approx 0.0417
*If playing doubles, your scores and opponents scores are averaged and then you figure out what the expected result is. Your match score is determined by how much you over or underperform against this.
*Your new dynamic rating is your newest match score averaged with your last three dynamic ratings.

Thanks, this is so helpful!

 

[TennisOTM]

Thanks, from others and other items I've read, I think both players pre-match ratings count so that USTA gives an "expected result" of the match. So, if I'm a 2.75 and she is a 3.19, the expected result would be 6-0 6-0 given we're so far apart to begin with. Getting 6 games to her 10 would beat the expected result, but then I'm lost as to how the match is calculated.

Your own pre-match rating only comes in when the match rating is averaged in. Your match rating from that match was probably something close to 3.09 (about 0.1 less than your opponents 3.19). So when that gets averaged in to your prior 2.75, it will make the average go up because you did better than expected.

If another player with a 3.30 rating played your opponent that day instead of you, and lost by the same score, their match rating would be the same 3.09, but that would have the opposite effect when averaged in, because she did worse than expected.

 

[nyta2]

Sadly and embarrassing to admit, I once was a 4.0 (when I was 18) and played D3 tennis. I quit exactly then and picked up the racket again three years ago (33 year break) and I finally have my strokes back. So, yes, I have those shots (except spin serves prob 8/10 vs. 10/10). I have a very good deep topspin forehand and my greatest strenght is my slice backhand and drop shots. I was finally picked up to play with high 3.5s at my club. But only because a few played with me and told others my game was better than my rating. BUT, I can't close tournament matches against 3.5s. I lose in a lot of tiebreakers (like close to 1/2). I keep playing up though in tournaments and I keep trying. But, if I just knew how close I was, it would help me immensely. It's a lot of time and $ to travel to tournaments so I'd like to know I'm making progress. Appreciate your response, I'm not giving up

ah, so probably a movement/timing issue.
if i were you, i'd play every pusher you know, and beat them first.
"playing up" can be deceiving, as you can block back hard hitters, and think you're making progress (and hoping the hard hitter, over hits)... but in reality you don't possess the skills to finish out points (eg. high & deep, aggressive deep midcourt, approach, volley/oh)... even if you beat the basher, it could be they beat themselves... so when you go back to the 3.5 pusher/moonballer (and everyone at 3.5 m/w are moonballer/pushers - else they are bashers that make mistakes), they beat you because now you have to generate pace, move, etc...

story time... when i took a 7-8 hiatus as a 4.5 top pursue other interests, when i returned i lost 5-7 to 4.0 pusher... was missing my "normal" attacking shots, then started pushing to try to win (which won't usually work at mid-high 4.5 anyway).... trained to get my footwork & timing down for the summer (basically called all the folks i used to beat, and worked my way up from bottom of my "tennis totem pole", back to the top, and returned to beat the 4.0 pusher 2,2...

moral of the story, is don't waste your time "playing up" if you don't even have the footwork/timing to beat a 3.5. it's good to occasionally "play up" to see what you need to do... and obviously good to practice with folks better than you if you can find them, but IMO, not worth traveling (time&$) to find much better players, unless you're already regularly beating all the best local players (yes, those pesky/annoying but hyper consistent pushers/moonballers).

 

[schmke]

Years back there was a lot of threads with specific information on the calculations of ratings, broke down by match differentials for game (from 0,0 to 6,6) tailing ratings, dubs, and more. That might have even been pre Schmke?

There have been some tables show up on the internet here and there from years ago. Presumably they had some truth to them at the time, do they still? Perhaps we'll never know.

FWIW I started doing my ratings around 2010.

 

[silverwyvern4]

Thanks, from others and other items I've read, I think both players pre-match ratings count so that USTA gives an "expected result" of the match. So, if I'm a 2.75 and she is a 3.19, the expected result would be 6-0 6-0 given we're so far apart to begin with. Getting 6 games to her 10 would beat the expected result, but then I'm lost as to how the match is calculated.

The most recent match counts the most, so if you're trying to get bumped up, in October keep playing until you get a really good match result and make that your final match before year end ratings. I believe the cutoff is November 1 or 15th?

 

[loveallcats]

Years back there was a lot of threads with specific information on the calculations of ratings, broke down by match differentials for game (from 0,0 to 6,6) tailing ratings, dubs, and more. That might have even been pre Schmke?

Any idea where that was? In forums? Wonder if we could find it on those internet archive sites.

 

[loveallcats]

The most recent match counts the most, so if you're trying to get bumped up, in October keep playing until you get a really good match result and make that your final match before year end ratings. I believe the cutoff is November 1 or 15th?

I thought that your rating is made up of your most recent match result averaged with your last 3 dynamic ratings. Does that mean that it counts for more than the others? I heard cutoff is around 11/1

 

[ChaelAZ]

Any idea where that was? In forums? Wonder if we could find it on those internet archive sites.

It was really early, so like maybe 07 or so. Or maybe later. It was one guy who started working though it but a lot of members chimed in to get a decent understanding of it. I am sure if you search ratings and look at the older threads it will be in there. Doubt they deleted it.

 

[TennisOTM]

The table below should be a decent start. I think there are a few possible game scores missing. The "match rating value" is what Tennisrecord adds to (if you won more games) or substracts from (if you lost more games) your opponent's pre-match rating to get your own match rating for that match.

Sometimes the add/subtract value differs by 0.01 from this table, which is probably due to rounding error - TR calculates the ratings to more than two decimals places but then rounds it off to two for the website, so we can't for sure know the precise formula I think.

I also saw some larger disagreements from what this table says for certain matches, and those seem to happen when the pre-match rating difference between the opponents was large (like around 0.5 or more). So they might use a different formula when that happens?

games won	games lost	game win %	match rating value
10​	10​	50.0%​	0​
11​	11​	50.0%​	0​
13​	12​	52.0%​	0.03​
12​	11​	52.2%​	0.03​
11​	10​	52.4%​	0.03​
10​	9​	52.6%​	0.03​
8​	7​	53.3%​	0.03​
13​	11​	54.2%​	0.05​
11​	9​	55.0%​	0.05​
13​	10​	56.5%​	0.06​
12​	9​	57.1%​	0.07​
11​	8​	57.9%​	0.09​
12​	8​	60.0%​	0.1​
13​	8​	61.9%​	0.13​
12​	7​	63.2%​	0.15​
11​	6​	64.7%​	0.17​
13​	7​	65.0%​	0.17​
12​	6​	66.7%​	0.2​
12​	5​	70.6%​	0.24​
12​	4​	75.0%​	0.29​
12​	3​	80.0%​	0.33​
12​	2​	85.7%​	0.38​
12​	1​	92.3%​	0.43​
12​	0​	100.0%​	0.48​

 

[silverwyvern4]

Does anyone know what shadowy cabal is behind the Tennis Record website?

 

[travlerajm]

Does anyone know what shadowy cabal is behind the Tennis Record website?

Someone who was doing a lot of winning and wanted a way to track it.

 

[silverwyvern4]

Oh no tennis record froze midupdate this week! Were they hit with a ransomware attack? I'd be willing to contribute 20 dollars toward the ransom!

 

[Icsa]

Oh no tennis record froze midupdate this week! Were they hit with a ransomware attack? I'd be willing to contribute 20 dollars toward the ransom!

It is summer. The owner is probably on a beach somewhere this week.

 

[silverwyvern4]

If my wife is trying to tank back to 3.0 so we can play 6.0 mixed doubles together again, is 0-6, RETIRE just as good as 0-6 0-6? If she only has to lose one set to lose maximum rating, that would save a lot of time and suffering for everyone on the court.

 

[Icsa]

If my wife is trying to tank back to 3.0 so we can play 6.0 mixed doubles together again, is 0-6, RETIRE just as good as 0-6 0-6? If she only has to lose one set to lose maximum rating, that would save a lot of time and suffering for everyone on the court.

You should respect everybody's time. You could tell your wife to hit everything really hard, mostly out. This way she can practice powerful shots while the other team will try to "defend" them. However, I hope you know that mixed matches matter for the rating only if she played only mixed matches the entire year. Otherwise mixed matches are ignored for the annual rating.

 

[Moon Shooter]

They very well may now. What I shared occurred several years ago.

And what I've seen Heather say in other webinars, and shared with me directly when I've talked with her, is that she got complaints from players saying their captain didn't play them because TLS/TR said they were a weak 3.5 (say 3.10) when in fact Heather looked up their actual dynamic rating and it was a high 3.5 (say 3.40). She cited the unreliability of the ratings causing captains to make bad line-up decisions as one of the negative effects of captains using TLS/TR.

It is interesting that she suggests using these more transparent rating systems leads to "bad" line-up decisions - based on her non-transparent system.

That said TR does do some very odd things with ratings from time to time.

 

[travlerajm]

It is interesting that she suggests using these more transparent rating systems leads to "bad" line-up decisions - based on her non-transparent system.

That said TR does do some very odd things with ratings from time to time.

The nice thing about TR is that it’s so transparent that the cause of any apparent anomaly can easily be figured out.

Most often, a player may have very different level in singles vs. doubles, and since these get lumped together, it could potentially lead a captain to make a poor line-up decision if he/she doesn’t look carefully enough to separate out the singles/doubles averages.

 

[Moveforwardalways]

Also, TR will almost always give most of the credit for a win to the partner with the higher rating. So if you are playing with someone who is improving quickly or is underrated for some reason, it can generate very high match ratings for you. I am not sure if USTA does the same thing. This also means that the lower rated partner will stay static or improve their rating only slightly. So it is really difficult to increase your TR rating while playing with a higher rated partner.

 

[travlerajm]

Also, TR will almost always give most of the credit for a win to the partner with the higher rating. So if you are playing with someone who is improving quickly or is underrated for some reason, it can generate very high match ratings for you. I am not sure if USTA does the same thing. This also means that the lower rated partner will stay static or improve their rating only slightly. So it is really difficult to increase your TR rating while playing with a higher rated partner.

I don’t believe that’s true.

TR gives equal credit or equal blame to both partners for the difference between team match rating and pre-match team expectation.

For example, if a 4.3 pairs with a 3.8, the team is expected to perform as an 8.1 level team.

If the team records an 8.4 match rating, that’s 0.3 above expectation, and the overperformance result gets shared equally.

The 4.3 gets a 4.45 match rating, and the 3.8 gets a 3.95 match rating.

The fact that the stronger player got more credit of total team ntrp is logical. But both players actually got equal credit for the above-expectation result.

Both players improved their ratings by the same amount.

 

[Moveforwardalways]

I don’t believe that’s true.

TR gives equal credit or equal blame to both partners for the difference between team match rating and pre-match team expectation.

For example, if a 4.3 pairs with a 3.8, the team is expected to perform as an 8.1 level team.

If the team records an 8.4 match rating, that’s 0.3 above expectation, and the overperformance result gets shared equally.

The 4.3 gets a 4.45 match rating, and the 3.8 gets a 3.95 match rating.

The fact that the stronger player got more credit of total team ntrp is logical. But both players actually got equal credit for the above-expectation result.

Both players improved their ratings by the same amount.

I have not seen it work that way in my matches

 

[travlerajm]

I have not seen it work that way in my matches

I only play mixed 8.0. My rating is usually about 4.4 and my partners are usually about 3.1. I will look more closely at my own record and confirm.

 

[travlerajm]

I only play mixed 8.0. My rating is usually about 4.4 and my partners are usually about 3.1. I will look more closely at my own record and confirm.

I just looked over my last several matches. TR worked exactly like I described in those.

In my last match, both my 3.5 partner and I received a match rating 0.05 above our pre-match ratings. So we both improved our rating equal amounts.

 

[Pspielha]

I just looked over my last several matches. TR worked exactly like I described in those.

In my last match, both my 3.5 partner and I received a match rating 0.05 above our pre-match ratings. So we both improved our rating equal amounts.

Also, TR will almost always give most of the credit for a win to the partner with the higher rating. So if you are playing with someone who is improving quickly or is underrated for some reason, it can generate very high match ratings for you. I am not sure if USTA does the same thing. This also means that the lower rated partner will stay static or improve their rating only slightly. So it is really difficult to increase your TR rating while playing with a higher rated partner.

“The difference in ratings of the members of a doubles team is held constant in a calculation of an individual match. If the two players are three hundredths (.03) of a point apart going into the match then they are three hundredths (.03) apart after the calculation in Step #4. However, once that number is averaged with the three previous dynamic ratings (Step #5) that difference may change. This is how we measure the performance of players as they change partners.”

http://assets.usta.com/assets/1/15/How%20Dynamic%20Ratings%20are%20Calculated%20for%20a%20Specific%20Match1.PDF

 

[loveallcats]

Slightly off topic - TR hasn't updated for a few weeks for mixed at least. I see some tournaments showing up but none of the mixed matches from last 3 weeks!

 

[silverwyvern4]

Slightly off topic - TR hasn't updated for a few weeks for mixed at least. I see some tournaments showing up but none of the mixed matches from last 3 weeks!

2024 or 2023 season?

 

[Bodhi312]

I see some mixed, no adult since August

 

[TennisOTM]

Slightly off topic - TR hasn't updated for a few weeks for mixed at least. I see some tournaments showing up but none of the mixed matches from last 3 weeks!

They are keeping up with the leagues I'm playing in - both mixed and adult. Must be something about the specific league(s) you are looking at.

 

[travlerajm]

My TR hasn’t updated for a month. There were always 3 things I could trust: death, taxes, and TR updating weekly. Now it feels like the world as I knew it is eroding.

 

[loveallcats]

My TR hasn’t updated for a month. There were always 3 things I could trust: death, taxes, and TR updating weekly. Now it feels like the world as I knew it is eroding.

Weird - why are some people's updating and others not?
TR was always more timely to import results than UTR, but now UTR seems to be back on track and TR is slipping!

 

[travlerajm]

Weird - why are some people's updating and others not?
TR was always more timely to import results than UTR, but now UTR seems to be back on track and TR is slipping!

To its credit UTR updates daily, but my both my doubles and my singles ratings jump around between UTR 4 and UTR 9 so it’s kind of meaningless.

 

[Icsa]

Weird - why are some people's updating and others not?
TR was always more timely to import results than UTR, but now UTR seems to be back on track and TR is slipping!

None of the 2024 leagues have been added: https://www.tennisrecord.com/adult/league/index.aspx
However I see some recent "Adult Other" matches showing up.
Likely USTA changed their formats again and TR didn't get a chance to make the update.

 

[Moon Shooter]

“The difference in ratings of the members of a doubles team is held constant in a calculation of an individual match. If the two players are three hundredths (.03) of a point apart going into the match then they are three hundredths (.03) apart after the calculation in Step #4. However, once that number is averaged with the three previous dynamic ratings (Step #5) that difference may change. This is how we measure the performance of players as they change partners.”

http://assets.usta.com/assets/1/15/How%20Dynamic%20Ratings%20are%20Calculated%20for%20a%20Specific%20Match1.PDF

Ok this might explain some of the odd things on TR. For example TR might have some peoples ratings inflated compared to NTRP. TR thinks they will get bumped but they don't. So for example in TR's system someone may have been a 3.22 but didn't get a bump to 3.5. Well then that player gets locked at 3.00 and the last 3 matches will likely average to 3.0. But then he plays with a Self rate player of his same actual strength. and lets say they play two 3.5 players that tr also rates at 3.32 or something. That new self rate ends up with a 3.44 and the other player that didn't get bumped ends up with a 3.07 or something.