NTRP and Mixed League

[TennisOTM]

Do you see what match ratings people get when they play mixed doubles that end in a .5? So for example mixed 6.5, 7.5, or 8.5?

It never lists it when I looked. I though it was because USTA does not include any of those leagues even in your mixed exclusive rating so TR also doesn't use those matches in their Mixed exclusive ratings calculations.

TR has match ratings for the Intermountain mixed X.5 league matches. I believe USTA will include these matches in your mixed exclusive rating as long as your local area / section decides to designate them rating-eligible.

TR also produces match ratings for matches that are not rating-eligible, but will color those ones green and presumably doesn't include them in the averaged ratings they calculate. I've only seen those for same-gender leagues, so I can't confirm if there are "green" mixed matches on TR somewhere.

 

[Klitz]

After reading through this thread, and USTAs website, I am still confused about Mixed matches impact on NTRP.

If someone ONLY plays Mixed USTA matches, is it possible for them to be bumped by USTA to the next rating from their initial self rating?

 

[travlerajm]

After reading through this thread, and USTAs website, I am still confused about Mixed matches impact on NTRP.

If someone ONLY plays Mixed USTA matches, is it possible for them to be bumped by USTA to the next rating from their initial self rating?

Yes. If you only play mixed, you can get a mixed exclusive rating. And you can be bumped up or down.

 

[Klitz]

Yes. If you only play mixed, you can get a mixed exclusive rating. And you can be bumped up or down.

OK, so people in the thread where saying that someone can circumvent this by posting/tanking 3 same-sex matches?

That would imply that the 3 same-sex matches carry INFINITE weight against the Mixed matches? Am I interpreting this right?

Also, I am still somewhat confused, because USTA site goes on and on about dynamic rating... Explicitly says Mixed only can't be dynamically rated. Yet, supposedly someone can still get bumped??? Why do they place so much emphasis on dynamic rating when it appears that it is entirely unnecessary with respect to getting bumped?

 

[Creighton]

OK, so people in the thread where saying that someone can circumvent this by posting/tanking 3 same-sex matches?

That would imply that the 3 same-sex matches carry INFINITE weight against the Mixed matches? Am I interpreting this right?

Also, I am still somewhat confused, because USTA site goes on and on about dynamic rating... Explicitly says Mixed only can't be dynamically rated. Yet, supposedly someone can still get bumped??? Why do they place so much emphasis on dynamic rating when it appears that it is entirely unnecessary with respect to getting bumped?

The dynamic rating generates the strikes to get someone dynamically disqualified during the season.

Yes, the same sex matches carry infinite more value in rating.

 

[Klitz]

The dynamic rating generates the strikes to get someone dynamically disqualified during the season.

Yes, the same sex matches carry infinite more value in rating.

Holy ****.

USTA engineered the most optimal mechanism to excel in Mixed into the rules.... Well, I guess if you only had to play 1 match instead of 3, that would be easier.

 

[leech]

Also, I am still somewhat confused, because USTA site goes on and on about dynamic rating... Explicitly says Mixed only can't be dynamically rated. Yet, supposedly someone can still get bumped??? Why do they place so much emphasis on dynamic rating when it appears that it is entirely unnecessary with respect to getting bumped?

You can't get dynamically DQ'd when you have anything other than an S or A rating. C, M, and T-rated players are not subject to strikes and dynamic DQs, but obviously can still get bumped up or down at year end.

 

[Creighton]

You can't get dynamically DQ'd when you have anything other than an S or A rating. C, M, and T-rated players are not subject to strikes and dynamic DQs, but obviously can still get bumped up or down at year end.

If you have a M or T rating you have to self rate to play adult league.

 

[Klitz]

If you have a M or T rating you have to self rate to play adult league.

Folks on here bash "self-rates" all day, but someone who only plays Mixed for a decade has to self-rate every three years, correct?

Even though they have USTA match history, they would be viewed as subhuman if they decided to join a men's league?

 

[Moon Shooter]

Folks on here bash "self-rates" all day, but someone who only plays Mixed for a decade has to self-rate every three years, correct?

Even though they have USTA match history, they would be viewed as subhuman if they decided to join a men's league?

This is what I think happens at least in my area:

If they only play mixed that ends with a .5 I am not sure if they ever will have to self rate again if they play over 3 matches every year. .5 mixed does not advance to nationals so USTA never adjusts peoples ratings based on those matches. Maybe they have to self rate again after 3 years even if they play mixed matches, but I am not sure.

If they play 3 or more mixed matches that ends in .0 then they will get a mixed exclusive and I think the mixed exclusive rating at year end. It will then be applied for mixed games so they do not need to self rate.

 

[schmke]

Folks on here bash "self-rates" all day, but someone who only plays Mixed for a decade has to self-rate every three years, correct?

Even though they have USTA match history, they would be viewed as subhuman if they decided to join a men's league?

If you have an M rating, you do not need to self-rate again to play Mixed. But if you have an M rating you have to self-rate to play Adult.

This seems to be the USTA saying M ratings don't really transfer to Adult play, at least completely. When an M self-rates, they must self-rate no lower than their current M level.

 

[Moon Shooter]

If you have an M rating, you do not need to self-rate again to play Mixed. But if you have an M rating you have to self-rate to play Adult.

This seems to be the USTA saying M ratings don't really transfer to Adult play, at least completely. When an M self-rates, they must self-rate no lower than their current M level.

What is interesting is if mixed ratings don't transfer to adult ratings well you would think that would mean adult ratings don't transfer to mixed ratings either. But they have adult ratings trump mixed ratings even for mixed leagues!

Do you find that M ratings transfer well to Adult play? Can you isolate any issues? As I indicated before, I think a male that plays with a lower rated female will tend to have a somewhat inflated mixed rating compared to adult rating. But a male that plays with a higher rated female will tend to have a deflated mixed rating compared to adult rating.

 

[Klitz]

If you have an M rating, you do not need to self-rate again to play Mixed. But if you have an M rating you have to self-rate to play Adult.

This seems to be the USTA saying M ratings don't really transfer to Adult play, at least completely. When an M self-rates, they must self-rate no lower than their current M level.

Have you found that M rated players are the least predictable for determining outcomes in your model?

 

[travlerajm]

Folks on here bash "self-rates" all day, but someone who only plays Mixed for a decade has to self-rate every three years, correct?

Even though they have USTA match history, they would be viewed as subhuman if they decided to join a men's league?

That’s me. But when I tried self-rating down a level just to see what would happen, it was auto-denied.

 

[TennisOTM]

I compared the Tennisrecord ratings for men who play in one 7.0 mixed league - just the guys who also play in men's leagues (at least 3 matches each). I tallied their latest averaged rating from men's matches (black on TR) vs. mixed matches (red on TR). Only 50 guys total, but the averages came out remarkably similar across the board:

Average ratings over all 50 players: 3.23 men, 3.23 mixed
Over 13 4.0 players: 3.67 men, 3.68 mixed
Over 23 3.5 players: 3.23 men, 3.22 mixed
Over 14 3.0 players: 2.80 men, 2.81 mixed

So at least for this group, the mixed ratings did quite well on average at mirroring results from men's league, no matter if they're playing with women who are the same, higher, or lower level.

Maybe a larger, more detailed analysis would find some interesting differences that this one didn't, but I'm guessing they'd be small.

 

[travlerajm]

I compared the Tennisrecord ratings for men who play in one 7.0 mixed league - just the guys who also play in men's leagues (at least 3 matches each). I tallied their latest averaged rating from men's matches (black on TR) vs. mixed matches (red on TR). Only 50 guys total, but the averages came out remarkably similar across the board:

Average ratings over all 50 players: 3.23 men, 3.23 mixed
Over 13 4.0 players: 3.67 men, 3.68 mixed
Over 23 3.5 players: 3.23 men, 3.22 mixed
Over 14 3.0 players: 2.80 men, 2.81 mixed

So at least for this group, the mixed ratings did quite well on average at mirroring results from men's league, no matter if they're playing with women who are the same, higher, or lower level.

Maybe a larger, more detailed analysis would find some interesting differences that this one didn't, but I'm guessing they'd be small.

Hypothesis:
If you take those same 4.0 players who played 7.0 mixed, and compared their 8.0 match ratings to their 7.0 match ratings, I expect they will have better ratings in 7.0 than in 8.0. I expect that their ratings would be 7.0 mixed > 4.0 men > 8.0 mixed.

In other words, their men’s rating equals their mixed rating because the “inflated” 7.0 results get canceled out by “deflated” 8.0 results.

 

[Moon Shooter]

I compared the Tennisrecord ratings for men who play in one 7.0 mixed league - just the guys who also play in men's leagues (at least 3 matches each). I tallied their latest averaged rating from men's matches (black on TR) vs. mixed matches (red on TR). Only 50 guys total, but the averages came out remarkably similar across the board:

Average ratings over all 50 players: 3.23 men, 3.23 mixed
Over 13 4.0 players: 3.67 men, 3.68 mixed
Over 23 3.5 players: 3.23 men, 3.22 mixed
Over 14 3.0 players: 2.80 men, 2.81 mixed

So at least for this group, the mixed ratings did quite well on average at mirroring results from men's league, no matter if they're playing with women who are the same, higher, or lower level.

did you actually see if if playing with a higher or lower rated female made a difference? Because you did not post the results showing that.
I predict that the mixed rating will be higher if the male player has a higher rating than the female player especially when they play against teams where the male does not have the higher rating.

 

[TennisOTM]

did you actually see if if playing with a higher or lower rated female made a difference? Because you did not post the results showing that.
I predict that the mixed rating will be higher if the male player has a higher rating than the female player especially when they play against teams where the male does not have the higher rating.

Well these are all 7.0 league players so the 4.0 men partnered with 3.0 women and the 3.0 men partnered with 4.0 women, and in both cases the men's mixed ratings matched their own ratings from men's league matches on average. Yes it would be interesting to test whether the opponent configuration makes a difference, though will probably need a larger pool of matches to get a reasonable sample size.

 

[TennisOTM]

Hypothesis:
If you take those same 4.0 players who played 7.0 mixed, and compared their 8.0 match ratings to their 7.0 match ratings, I expect they will have better ratings in 7.0 than in 8.0. I expect that their ratings would be 7.0 mixed > 4.0 men > 8.0 mixed.

In other words, their men’s rating equals their mixed rating because the “inflated” 7.0 results get canceled out by “deflated” 8.0 results.

I think there are only a couple of these guys who also play 8.0, so even if they rated lower there I'm guessing those didn't deflate the overall average much, but I'll try to check when I get a chance.

 

[TennisOTM]

Hypothesis:
If you take those same 4.0 players who played 7.0 mixed, and compared their 8.0 match ratings to their 7.0 match ratings, I expect they will have better ratings in 7.0 than in 8.0. I expect that their ratings would be 7.0 mixed > 4.0 men > 8.0 mixed.

In other words, their men’s rating equals their mixed rating because the “inflated” 7.0 results get canceled out by “deflated” 8.0 results.

Only one of those 4.0 players played substantial matches at different mixed levels, but this guy is a treasure trove of data so I couldn't resist diving in. He plays a lot of men's 4.0 doubles and a TON of mixed, across 6 different levels including X.5's. Here's how his TR match ratings average out over the last few years.

Men's 4.0 (34 doubles matches): 3.81
Mixed Any (60 matches): 3.81

Mixed 7.0 (6 matches): 3.83
Mixed 7.5 (9 matches): 3.75
Mixed 8.0 (19 matches): 3.79
Mixed 8.5 (9 matches): 3.76
Mixed 9.0 (16 matches): 3.92
Mixed 9.5 (1 match): 3.89

Surprising? Your specific hypothesis was correct for this guy: 7.0 mixed > 4.0 men > 8.0 mixed, but those are too close to conclude a difference, and then the other levels don't fit the pattern. His best level appears to be 9.0, go figure.

 

[travlerajm]

Only one of those 4.0 players played substantial matches at different mixed levels, but this guy is a treasure trove of data so I couldn't resist diving in. He plays a lot of men's 4.0 doubles and a TON of mixed, across 6 different levels including X.5's. Here's how his TR match ratings average out over the last few years.

Men's 4.0 (34 doubles matches): 3.81
Mixed Any (60 matches): 3.81

Mixed 7.0 (6 matches): 3.83
Mixed 7.5 (9 matches): 3.75
Mixed 8.0 (19 matches): 3.79
Mixed 8.5 (9 matches): 3.76
Mixed 9.0 (16 matches): 3.92
Mixed 9.5 (1 match): 3.89

Surprising? Your specific hypothesis was correct for this guy: 7.0 mixed > 4.0 men > 8.0 mixed, but those are too close to conclude a difference, and then the other levels don't fit the pattern. His best level appears to be 9.0, go figure.

It actually fits @schmke ’s data well. And fits the rule that doubles is a strongest-link-wins sport. The two categories where he is part of a significantly imbalanced pairing and has a significant advantage are 9.0 (where a 5.0 gal partner is the best player on the court in 80% of the matches, as long as they don’t face a 5.0 guy); and to a lesser extent, 7.0, where he would have the advantage roughly 60% of the time, whenever he plays a 3.0 opposing guy. When he plays 8.0, he is at a disadvantage about 20% of the time when he faces a 4.5 guy, so he does slightly worse than in 7.0.

In other words, testable Hypothesis 2:

The ratings are a function of both the degree of imbalance in the on-court pairing (degree of advantage) and the fraction of matches where the advantage occurs.

A. I expect that this 4.0 treasure trove dude will have higher average rating in 7.0 matches where his opponents are 3.0M/4.0F or 3.5/3.5 (moderate advantage) than when he plays fellow 4.0M/3.0F teams.

B. I also expect that this same 4.0 dude will have higher average mixed rating when he plays fellow 4.0/4.0 teams than when he plays 4.5M/3.5F teams (huge disadvantage).

C. I also expect that this same 4.0 dude will have higher average rating in 9.0 matches when he faces 4.5/4.5 teams (moderate advantage) than when he faces 5.0M/4.0F teams (slight disadvantage).

D. I also expect that this same 4.0 dude will have average mixed rating in decreasing strength mirroring scheme’s data as follows:

1. 9.0 vs 4.5/4.5 (big advantage)
2. 7.0 vs 3.5/3.5 or 3.0M/4.0F (moderate advantage).
3. 7.0 vs 4.0M/3.0F and 8.0 vs 4.0/4.0 (equal teams)
4. 9.0 vs 5.0M/4.0F (slight disadvantage)
5. 8.0 vs 4.5M/3.5F (huge disadvantage)

 

[TennisOTM]

I took a stab at your nice hypotheses for this player. I excluded some matches where there was an under-level team (e.g. a 4.0/3.5 playing 8.0), and categorized the other ones. Here goes:

A. I expect that this 4.0 treasure trove dude will have higher average rating in 7.0 matches where his opponents are 3.0M/4.0F or 3.5/3.5 (moderate advantage) than when he plays fellow 4.0M/3.0F teams.

Not enough data to test this one for this guy - he faced 3.5/3.5 pairs in every match but one.

B. I also expect that this same 4.0 dude will have higher average mixed rating when he plays fellow 4.0/4.0 teams than when he plays 4.5M/3.5F teams (huge disadvantage).

This hypothesis isn't supported: he averaged 3.83 in 8 matches against 4.0/4.0 teams and 3.85 in 10 matches against 4.5M/3.5F teams. Only 3 matches against 3.5M/4.5W teams (3.80 average).

C. I also expect that this same 4.0 dude will have higher average rating in 9.0 matches when he faces 4.5/4.5 teams (moderate advantage) than when he faces 5.0M/4.0F teams (slight disadvantage).

He averaged 3.86 in 16 matches against 4.5/4.5 teams. Only found 1 match against a 5.0M/4.0F team (he rated 4.04 in that one). Three matches against a fellow 4.0M/5.0W team averaged 3.95.

D. I also expect that this same 4.0 dude will have average mixed rating in decreasing strength mirroring scheme’s data as follows:

1. 9.0 vs 4.5/4.5 (big advantage)
2. 7.0 vs 3.5/3.5 or 3.0M/4.0F (moderate advantage).
3. 7.0 vs 4.0M/3.0F and 8.0 vs 4.0/4.0 (equal teams)
4. 9.0 vs 5.0M/4.0F (slight disadvantage)
5. 8.0 vs 4.5M/3.5F (huge disadvantage)

Here are the averages:
1. 9.0 vs 4.5/4.5 (16 matches): 3.86
2. 7.0 vs 3.5/3.5 or 3.0M/4.0F (4 matches): 3.86
3. 7.0 vs 4.0M/3.0F and 8.0 vs 4.0/4.0 (9 matches): 3.79
4. 9.0 vs 5.0M/4.0F (1 match): 4.04
5. 8.0 vs 4.5M/3.5F (10 matches): 3.85

Not great sample sizes, but it does seem like he did a bit better in #'s 1&2 vs. #3. But then he also did well in #4&5 compared to #3.

Here's another way of comparing - all matches where the opposing male is the same level (4.0), higher level, or lower level than he is:

Any league vs. 4.0M/X.XW (27 matches): 3.78
Any league vs. 4.5 or 5.0M/X.XW (32 matches): 3.84
Any league vs. 3.5M/X.XW (8 matches): 3.83

Not really a big difference, but if those differences are real then perhaps he does press his advantage when he's the clear best player on the court, but on the flip side maybe he also steps his game up a notch when the opposing male is better.

Both things could be happening across a lot of different players and perhaps in combination they even everything out on average. Some players are good/bad at playing the role of the best player on the court, some are good/bad at holding their own as a weaker player. Some are good at both, bad at both, or good at one and bad at the other.

 

[Moon Shooter]

I compared the Tennisrecord ratings for men who play in one 7.0 mixed league - just the guys who also play in men's leagues (at least 3 matches each). I tallied their latest averaged rating from men's matches (black on TR) vs. mixed matches (red on TR). Only 50 guys total, but the averages came out remarkably similar across the board:

Average ratings over all 50 players: 3.23 men, 3.23 mixed
Over 13 4.0 players: 3.67 men, 3.68 mixed
Over 23 3.5 players: 3.23 men, 3.22 mixed
Over 14 3.0 players: 2.80 men, 2.81 mixed

So at least for this group, the mixed ratings did quite well on average at mirroring results from men's league, no matter if they're playing with women who are the same, higher, or lower level.

Maybe a larger, more detailed analysis would find some interesting differences that this one didn't, but I'm guessing they'd be small.

TennisOTM
Yes that data does cut against my prediction that the men would rate higher if they are paired with a lower level female.

I would note that we are still going by the assumption that the mixed rating algorithm only uses the male and female adult dynamic rating and adds them together and in no way modifies them. If they do some modifications in the mixed rating algorithm then they may be countering/accounting for the inflation/deflation I would predict.

It would be nice to know the opponents but I don't think there is a reason to assume they wouldn't generally be about the same spread as the people you tracked. So for the 13 4.0 men we might assume that their opponents were approximately 4.0 M and 3.0 F about 3 times and 3.5 male and female about 6 times and 4.0 female and 3.0 male about 4 times. And something similar for the teams with a 4.0 female and 3.0 male.

It is obviously a small sample size but I agree it is some evidence that cuts against my theory.

This data also cuts against the theory that tries to explain Schmke's data by saying unbalanced teams have an advantage. The difference in ratings is way to small to account for these results:

"For 7.0, here is how it breaks out for the 18 & over Mixed leagues over the past few years:

• 4.0/3.0 vs 3.5/3.5 - The 4.0/3.0 wins 60% of the time
• 3.0/4.0 vs 3.5/3.5 - The 3.0/4.0 wins 52% of the time
• 4.0/3.0 vs 3.0/4.0 - The 4.0/3.0 wins 57% of the time

What NTRP rating combinations works best in 18 & over Mixed? More Interesting Tennis League Stats

It is time for 18 & over Mixed teams to form in my area for the 2016 season. This is an early start league for us so we use our early sta...

computerratings.blogspot.com

To get in a situation where you are winning 57-60% of the time I would think you would need at least a .1 rating difference. Your data is showing they only get a .01 difference which is not even close to accounting for 57-60% winning chances.

That brings us to the other theory that might explain Schmke's data. That generally speaking mixed teams are trying to fill their courts with multiple players that are half the total number of levels allowed. So 7.0 teams are generally playing two 3.5 players. Your data tends to demonstrate that since there are more 3.5 pairs. And so when you do have a few of the 3.0 and 4.0 pairs they may tend to be closer to the top of their rating. And the 3.5s are likely more in the average since you need more of them to fill the courts. Your averages wont really show this. Because you averages just show the average ratings. So for example you say the 13 4.0 men average 3.67. That could be because 8 are 3.75ish and 5 are 3.54ish. They could all have the same performance rating at mixed but you would expect the 8 players that are 3.75ish to win especially if they can easily be paired up with some higher rated 3.0 females. This could easily result in a 57-60% winning result for the unbalanced teams simply because they tend to have people that are higher rated with in their level.

We would still have difficulty explaining why this doesn't work for 4.0 female and 3.0 male teams.

 

[Moon Shooter]

Only one of those 4.0 players played substantial matches at different mixed levels, but this guy is a treasure trove of data so I couldn't resist diving in. He plays a lot of men's 4.0 doubles and a TON of mixed, across 6 different levels including X.5's. Here's how his TR match ratings average out over the last few years.

Men's 4.0 (34 doubles matches): 3.81
Mixed Any (60 matches): 3.81

Mixed 7.0 (6 matches): 3.83
Mixed 7.5 (9 matches): 3.75
Mixed 8.0 (19 matches): 3.79
Mixed 8.5 (9 matches): 3.76
Mixed 9.0 (16 matches): 3.92
Mixed 9.5 (1 match): 3.89

Surprising? Your specific hypothesis was correct for this guy: 7.0 mixed > 4.0 men > 8.0 mixed, but those are too close to conclude a difference, and then the other levels don't fit the pattern. His best level appears to be 9.0, go figure.

As far as the individual 4.0 player, one player is obviously too small of a sample. He is also a very high rated 4.0. (3.81 would probably put him in the top 85% of 4.0 males) So for what little looking at his results is worth he cuts hard against the dominance theory.

Are you saying his mixed double performance rating in adult league with other men is a 3.81 and his performance in 7.0 mixed doubles presumably with another female 3.0 or lower is only a 3.83? 2 hundredths of a point is not at all close to explaining schmke's data where we expect him to win 57-60% of his games based on the dominance theory. And then the dip to 3.75 in 7.5 league compared to the bounce back up to 3.79 in 8.0 also runs counter to the dominance theory.

But overall I don't think examining one person is going to give us much that is useful.

 

[TennisOTM]

I would note that we are still going by the assumption that the mixed rating algorithm only uses the male and female adult dynamic rating and adds them together and in no way modifies them. If they do some modifications in the mixed rating algorithm then they may be countering/accounting for the inflation/deflation I would predict.

It's pretty easy to see that Tennisrecord is using the same algorithm for mixed and same-gender matches. I found two of my own matches that ended with the same exact score, one in a men's match and one in a mixed match, and the match ratings were calculated exactly the same: the sum of the partners' match ratings had the same (score-based) differential from the pre-match rating total of the opponents, while maintaining the pre-match differential of the partners.

That brings us to the other theory that might explain Schmke's data. That generally speaking mixed teams are trying to fill their courts with multiple players that are half the total number of levels allowed. So 7.0 teams are generally playing two 3.5 players. Your data tends to demonstrate that since there are more 3.5 pairs. And so when you do have a few of the 3.0 and 4.0 pairs they may tend to be closer to the top of their rating. And the 3.5s are likely more in the average since you need more of them to fill the courts. Your averages wont really show this. Because you averages just show the average ratings. So for example you say the 13 4.0 men average 3.67. That could be because 8 are 3.75ish and 5 are 3.54ish. They could all have the same performance rating at mixed but you would expect the 8 players that are 3.75ish to win especially if they can easily be paired up with some higher rated 3.0 females. This could easily result in a 57-60% winning result for the unbalanced teams simply because they tend to have people that are higher rated with in their level.

We would still have difficulty explaining why this doesn't work for 4.0 female and 3.0 male teams.

A good test would be how often were the 4.0/3.0 pairs favored to win against 3.5/3.5 or 3.0/4.0 pairs, based on having a higher sum of dynamic ratings going into the match. If they had a higher sum 50% of the time, but then actually won 60%, that would tell you that the straight sum approach might warrant some adjustment. My guess though is that 4.0/3.0 pairs (especially 4.0M/3.0F pairs) do tend to be favored >50% of the time, largely because the 3.0's who play 7.0 tend to be high-end 3.0's (especially women). This should be testable but might take a lot of work to gather the right data.

 

[Moon Shooter]

It's pretty easy to see that Tennisrecord is using the same algorithm for mixed and same-gender matches. I found two of my own matches that ended with the same exact score, one in a men's match and one in a mixed match, and the match ratings were calculated exactly the same: the sum of the partners' match ratings had the same (score-based) differential from the pre-match rating total of the opponents, while maintaining the pre-match differential of the partners.

"I tallied their latest averaged rating from men's matches (black on TR) vs. mixed matches (red on TR)."

Ok did you take the average performance rating for their mixed matches in each of the categories or just the ending "mixed match" rating for each of the players?

Because remember the mixed games do not effect their rating if someone also has an adult rating. So lets say someone is a 3.5 male with a 3.35 dynamic rating. Lets say In match 1 they get a match performance rating of 3.45. And that match should bump their rating to 3.39. So the next match (match 2) they should be going into the match with a 3.39 rating and if they get another 3.45 performance rating they may have their rating bumped to 3.42. So then match three they would have a dynamic rating of 3.42 going in. But I think the way usta (and therefore TR) does it is they are not going into the match 2 with a 3.39 rating and match 3 with a 3.42 rating. They are going into match 2 with a 3.35 rating and match 3 with a 3.35 rating because the mixed match matches don't count at all for people with an adult rating. So the actual end rating will of course end up showing they have a pretty close rating to where they started. The ratings keep resetting to the adult rating after every match.

If the person had a mixed exclusive rating and no adult rating then going into the match they would go into those next matches with a higher dynamic rating.

On the other hand, if you took the average performance rating for each match in each category (7.0, 7.5, 8.0, 8.5 etc) and averaged those, then I think you will have a better idea of whether people with a mixed exclusive rating will over perform rating wise. It still may be a bit skewed but I think it would capture what we are looking for pretty well.

A good test would be how often were the 4.0/3.0 pairs favored to win against 3.5/3.5 or 3.0/4.0 pairs, based on having a higher sum of dynamic ratings going into the match. If they had a higher sum 50% of the time, but then actually won 60%, that would tell you that the straight sum approach might warrant some adjustment. My guess though is that 4.0/3.0 pairs (especially 4.0M/3.0F pairs) do tend to be favored >50% of the time, largely because the 3.0's who play 7.0 tend to be high-end 3.0's (especially women). This should be testable but might take a lot of work to gather the right data.

Yes I think that would be a decent test. I think that if you take a team (team A) with a male with an adult rating of 3.95 and a female with an adult rating of 2.95 they are going to do better than, Team B with a male with an adult rating of 2.95 and a female of 3.95. I also think Team A will do better than team C that has a two 3.45 players. But I think team C would do just as well or better than team B.


I don't really buy the dominance of one player on the court I think schmke's data is explained by just the fact that there is more total skill on the court. I would think that teams that had similar UTRs (which uses the same ratings for men and women) for both partners do as well or better than teams that have large difference in UTRs. So take a team with two 5.3 UTR players and they will perform at least as well as a team with a 3.3 and a 7.3 UTR. I even think they will likely do better just because they have a clear strategy of who to hit the ball to. There is also a psychological aspect for both the weaker player seeing that they are losing the match for their team and the stronger player possibly getting frustrated that their partner makes what they consider basic errors in judgment or ability.

 

[TennisOTM]

Because remember the mixed games do not effect their rating if someone also has an adult rating. So lets say someone is a 3.5 male with a 3.35 dynamic rating. Lets say In match 1 they get a match performance rating of 3.45. And that match should bump their rating to 3.39. So the next match (match 2) they should be going into the match with a 3.39 rating and if they get another 3.45 performance rating they may have their rating bumped to 3.42. So then match three they would have a dynamic rating of 3.42 going in. But I think the way usta (and therefore TR) does it is they are not going into the match 2 with a 3.39 rating and match 3 with a 3.42 rating. They are going into match 2 with a 3.35 rating and match 3 with a 3.35 rating because the mixed match matches don't count at all for people with an adult rating. So the actual end rating will of course end up showing they have a pretty close rating to where they started. The ratings keep resetting to the adult rating after every match.

A check through TR shows that they don't calculate mixed ratings the way you describe. It appears that the only time that the mixed match calculations use someone's adult rating is when it's that players first mixed match. If that first-time mixed player has a current adult rating it will use that to calculate the match rating for their partner and opponents. But if all four players have prior mixed ratings, only those numbers are used and their current adult ratings are ignored.

 

[Moon Shooter]

A check through TR shows that they don't calculate mixed ratings the way you describe. It appears that the only time that the mixed match calculations use someone's adult rating is when it's that players first mixed match. If that first-time mixed player has a current adult rating it will use that to calculate the match rating for their partner and opponents. But if all four players have prior mixed ratings, only those numbers are used and their current adult ratings are ignored.

Wow that is interesting. Thanks for taking a look at these numbers.

Are you sure they don't discount a when the female players rating when she is higher rated or do something that would have that effect? The reason I ask is because it does seem to me that a mid 4.0 male is stronger than a mid 4.0 female and this UTR chart seems pretty accurate at charting out the differences:

https://i0.wp.com/blog.universaltennis.com/wp-content/uploads/2018/07/Player-Segments.png?resize=840%2C560&ssl=1

So a male mid 4.0 would be about a UTR 5.5-6.0 player and a female mid 4.0 would be about 4.0-4.5 UTR. A mid male 3.0 would be about UTR 3.5 and a mid 3.0 female would be a UTR 2.5-3.0. So the male 4.0 would be about 5.75 and the female 3.0 would be about 2.75.(so total 8.50 on the court) The female 4.0 would be about 4.25 and the male 3.0 would be about 3.5. (so total 7.75 on the court.) That difference of .75 is not huge but I would think it would be enough to have some identifiable effect.

I have a team with a few self rates that have played mixed doubles - myself included. I will try to see how their adult rating transfers. The one player that has played extensive mixed doubles and also has a adult rating has a .16 difference between the ratings with his mixed rating being higher. I think that is just because his adult rating is stale and didn't involve many games.

 

[gooberwho]

the tank is on for 4.0. he lost a 4.0 18-and-over singles match 0 and 0 over the weekend. suspicious? nah.

the tank is on for 4.5 despite his being suspended last year and missing 3 months. he lost to a self-rated 4.0 0 and 1 in singles. Guy is shameless.

 

[TennisOTM]

the tank is on for 4.5 despite his being suspended last year and missing 3 months. he lost to a self-rated 4.0 0 and 1 in singles. Guy is shameless.

He has the doubles UTR of an above-average 4.5, but the singles UTR of an average 3.5. Is there even a remote chance that such a difference in singles vs. doubles ability could be real??

Interesting that his men's team still won the match despite his blowout singles loss, and they're in first place after one match. If his team is actually good, would his captain tolerate playing him at singles again?

 

[FuzzyYellowBalls]

He has the doubles UTR of an above-average 4.5, but the singles UTR of an average 3.5. Is there even a remote chance that such a difference in singles vs. doubles ability could be real??

Interesting that his men's team still won the match despite his blowout singles loss, and they're in first place after one match. If his team is actually good, would his captain tolerate playing him at singles again?

Who is this, you can pm if you are so kind. It's not possible, low/mid 4.0 singles UTR and high 4.5 is possible, but a real 3.5 UTR in singles, that person would have at one point had a singles rating somewhere above that and have to tank 0-6 0-6 to get it to naturally move to a real 3.5 rating.

 

[!<-_->!]

Looks like that team is short on players so they may not care if he wins or loses

 

[Jonathyl]

He has the doubles UTR of an above-average 4.5, but the singles UTR of an average 3.5. Is there even a remote chance that such a difference in singles vs. doubles ability could be real??

Interesting that his men's team still won the match despite his blowout singles loss, and they're in first place after one match. If his team is actually good, would his captain tolerate playing him at singles again?

I've played him at Sectionals in mixed doubles (they won in a Supertie), and from my personal experience, there is no way his singles record is legit. He moves well, and bad movement would be the main difference to account for the discrepancy between ratings. He's got this wicked kick serve that most 4.0's would struggle returning, he's got fantastic net skills, his ground strokes aren't super powerful but very consistent (though he could have been holding back on power when we played). Imo he'd destroy 95% of 4.0's.

 

[onehandbh]

It actually fits @schmke ’s data well. And fits the rule that doubles is a strongest-link-wins sport.

Do you believe this also applies to ATP doubles teams?

 

[travlerajm]

Do you believe this also applies to ATP doubles teams?

Moreso in tournaments that use no-ad scoring, where if you have one stud on your team, he can break serve by himself.

 

[Jonathyl]

the tank is on for 4.5 despite his being suspended last year and missing 3 months. he lost to a self-rated 4.0 0 and 1 in singles. Guy is shameless.

And he recently lost 0-6, 0-6 this past weekend. To a guy who went 1-12 last year...

 

[Purestriker]

And he recently lost 0-6, 0-6 this past weekend. To a guy who went 1-12 last year...

Wow. What section is this?

 

[Jonathyl]

Wow. What section is this?

Norcal

 

[Purestriker]

Norcal

Interesting.

 

[TennisOTM]

And he recently lost 0-6, 0-6 this past weekend. To a guy who went 1-12 last year...

Meanwhile he is registered on six different mixed doubles teams and performing like a solid 4.5 player in those leagues. Will USTA "reward" him with a 4.0C next year if he keeps up this pattern?

Looking back at last year, I wonder if someone at USTA actually intervened to give him a 4.5C. His estimated ratings on TR from Adult leagues were WAY below the 4.0/4.5 border, but he got a 4.5C anyway. Perhaps the USTA has him flagged for special consideration, and his current "strategy" (if that's what it is) will be fruitless.

 

[Jonathyl]

Meanwhile he is registered on six different mixed doubles teams and performing like a solid 4.5 player in those leagues. Will USTA "reward" him with a 4.0C next year if he keeps up this pattern?

Looking back at last year, I wonder if someone at USTA actually intervened to give him a 4.5C. His estimated ratings on TR from Adult leagues were WAY below the 4.0/4.5 border, but he got a 4.5C anyway. Perhaps the USTA has him flagged for special consideration, and his current "strategy" (if that's what it is) will be fruitless.

I think he was manually moved up by USTA after his Tri-level team won. There's another guy that was 3.5 on his Tri-level team that went 1-4 at 3.5, and then something like 35-3 in in mixed and combo (from 6.0 mixed all the way to 8.5 combo) and he got bumped to 4.0.

 

[!<-_->!]

I think he was manually moved up by USTA after his Tri-level team won. There's another guy that was 3.5 on his Tri-level team that went 1-4 at 3.5, and then something like 35-3 in in mixed and combo (from 6.0 mixed all the way to 8.5 combo) and he got bumped to 4.0.

This the same guy who lost his first mixed 9.0 match?

 

[Jonathyl]

This the same guy who lost his first mixed 9.0 match?

Indeed

 

[!<-_->!]

Indeed

Can quite confidently say this fellow was intentionally throwing 3.5 matches. Looks like the pupil met the same fate as the master though with the bump.

 

[PK6]

Good to see these guys getting bumped up instead of down. The 3.0s has killed USTA. This is why USTA needs to get rid of self ratings!!! As I said if you’ve played high school or college tennis your automatic 4.0s

 

[Jonathyl]

Good to see these guys getting bumped up instead of down. The 3.0s has killed USTA. This is why USTA needs to get rid of self ratings!!! As I said if you’ve played high school or college tennis your automatic 4.0s

There's another guy on their teams who was a 3.5 who regularly goes undefeated in mixed 6.0 to 8.0 (going 10-0, 9-0, etc) who last year went 0-3 at 3.5, getting bageled in every match. He's the team captain btw. USTA just rewarded his hard efforts with a 3.0A.

He's been tanking for over 10 years. I guess he's just smart enough not to join their tri-level teams

 

[PK6]

There's another guy on their teams who was a 3.5 who regularly goes undefeated in mixed 6.0 to 8.0 (going 10-0, 9-0, etc) who last year went 0-3 at 3.5, getting bageled in every match. He's the team captain btw. USTA just rewarded his hard efforts with a 3.0A.

He's been tanking for over 10 years. I guess he's just smart enough not to join their tri-level teams

That’s one of reasons why I quit playing tennis. Whole system is a joke-needs to be blown up!!!

 

[Moon Shooter]

There's another guy on their teams who was a 3.5 who regularly goes undefeated in mixed 6.0 to 8.0 (going 10-0, 9-0, etc) who last year went 0-3 at 3.5, getting bageled in every match. He's the team captain btw. USTA just rewarded his hard efforts with a 3.0A.

He's been tanking for over 10 years. I guess he's just smart enough not to join their tri-level teams

I wouldn't say someone is "smart" just because they see this hole big enough to drive a truck through in how USTA treats Mixed matches.

Of course, this will happen as long as they do not include mixed matches in your rating used for mixed matches. There is plenty USTA could do to correct this abuse but they simply don't care.

 

[Creighton]

I wouldn't say someone is "smart" just because they see this hole big enough to drive a truck through in how USTA treats Mixed matches.

Of course, this will happen as long as they do not include mixed matches in your rating used for mixed matches. There is plenty USTA could do to correct this abuse but they simply don't care.

The problem is if you include mixed matches, people will just throw mixed matches to save their same gender leagues.

I would rather people cheat in mixed than the same gender leagues. It's the lesser of the evils.

 

[TennisOTM]

I think he was manually moved up by USTA after his Tri-level team won. There's another guy that was 3.5 on his Tri-level team that went 1-4 at 3.5, and then something like 35-3 in in mixed and combo (from 6.0 mixed all the way to 8.5 combo) and he got bumped to 4.0.

There's another guy on their teams who was a 3.5 who regularly goes undefeated in mixed 6.0 to 8.0 (going 10-0, 9-0, etc) who last year went 0-3 at 3.5, getting bageled in every match. He's the team captain btw. USTA just rewarded his hard efforts with a 3.0A.

He's been tanking for over 10 years. I guess he's just smart enough not to join their tri-level teams

Interesting. It looks like Tri-level and men's combo are not supposed to count for ratings in NorCal (at least TR does not think they count). So if USTA bumped them up manually based on results from those leagues it would seem to go against their written policy. Maybe they only do that in specific cases where players are on some kind of prior-offender or grievance list.

Then the next question is whether mixed league results alone would be enough for USTA to make a similar "manual" bump-up decision? Based on your second example, maybe not.