USTA hints at singles and doubles specific NTRP ratings

[schmke]

I received a USTA newsletter e-mail today and it had a link to a story on what the NTRP is all about (http://www.usta.com/Adult-Tennis/US...p_ratings_continue_to_drive_usta_league_play/). Not much substance there, just a little history on how it started and self-rating was introduced, then a few stats on bumps. But it ends with this:

Jones said the next step in the NTRP’s evolution would be to make rating differentiations between singles and doubles results.

“I’d say about 20 percent of players have a significant difference in their singles and doubles ability, either because of age or hand-eye coordination,” he explained. “It would be great to see that ratings change happen in the coming years.”​

This is certainly something I think makes sense, I've written about it before (http://computerratings.blogspot.com/2014/03/should-there-be-separate-ntrp-ratings.html and http://computerratings.blogspot.com/2015/03/is-this-usta-league-player-tanking-to.html), but I had not seen anything from the USTA hinting that they agreed or would be open to doing it.

Now, it introduces a lot of complications as what do you do if a player has different singles and doubles ratings? What teams can they sign-up for and presumably they become limited to just playing singles or doubles on different teams?

So it is interesting to think about, but I'd still be surprised to see the USTA do it.

 

[CASmash]

I think this would be good. I know so many players that are vastly better at one than the other.

 

[pennc94]

Great - now they need sex and age groups specific ratings and then we're all set.

 

[chay337]

Good, and use to higher rating as the one they use to join a specific level. Ex. If Singles rating is 3.0 and doubles rating is 3.5, they can join a 3.5 level team or one level higher (3.5-4.0).

Actually, what's the point? Unless that allows a person can join a 3.0 team to just play singles.

 

[Topaz]

This would make a lot of sense, however I'd be surprised if it ever happened. I'm pretty darned sure it was my singles that got my bumped down last year...my other scores were competitive at level. It has, of course, worked out to my advantage this year, but really isn't accurate.

 

[g4driver]

I read that newsletter this am also. My thoughts were you could easily have a player with dual rating join one team, say a 3.5 men's team with a S3.5/D4.0 Rating. The first number being the singles rating and the second the doubles rating, on a printout and by the players name. Putting the "S" or "D" before the number indicates Singles or Doubles, as opposed to an "S" for Self-Rated, or "D" for Disqualified after the number.

Clean? Not as clean as it could be. Maybe "SNG 3.5 / DBL 4.0"

In a 3.5 match, this player could only play singles on a 3.5 team and never 3.5 doubles. If the Captain put him at 3.5 Doubles, and he plays one point in Doubles, that court is simply entered as a default and the Captain gets smarter the next time.

If the player never plays singles or never plays doubles the rating would simply be a single "3.5" which means the ratings for singles and doubles are the same. He could appear as Joe Bagodonuts 3.5, meaning Joe's ratings for singles and doubles are the same. The dual rating folks would stick out like a sore thumb. e.g., Sally Smith SNG 3.5/DBL 4.0

The mistake IMO would be to put both numbers by every player's name. If you only have one rating for both singles and doubles, by only having one number by your name, makes those dual rating easier to spot to both teams and especially to both Captains.

This wouldn't be that difficult to implement as I know most players play singles or doubles, not both. On my Men's 4.0 team, only four out of 17 guys played both singles and doubles. One guy only played singles. The remaining 12 guys were doubles only. Just off of that team, only four of the 17 guys would have generated the dual number by their name. The singles and doubles only guys having just one number by their name would mean they have only have one rating 4.0, until such time they played enough singles or doubles matches to generate a dual rating.


This doesn't have to be hard, but I am have faith the USTA can make it more difficult than needed.

 

[Topaz]

G4...lots of the women I know who play singles also play doubles, just because there is so much more doubles to be played!

 

[g4driver]

Topaz,

I never implied or meant to imply that men or women can't play bought, but rather it is more rare to play both on a team during a season. We had four guys who played both out of 17 guys.

I much prefer doubles and am better at doubles, but I played six singles matches this spring and only four doubles matches on two men's team. There are plenty of players who can play both singles and doubles but are only used as one or the other during a season.

 

[sureshs]

The tours have separate rankings for doubles

 

[Topaz]

Yes, I see what you mean...I was thinking more of the whole season overall. We actually have a singles league, though it is short, that was started mostly because the singles players just wanted more opportunities to play singles!

 

[Nostradamus]

I received a USTA newsletter e-mail today and it had a link to a story on what the NTRP is all about (http://www.usta.com/Adult-Tennis/US...p_ratings_continue_to_drive_usta_league_play/). Not much substance there, just a little history on how it started and self-rating was introduced, then a few stats on bumps. But it ends with this:

Jones said the next step in the NTRP’s evolution would be to make rating differentiations between singles and doubles results.

“I’d say about 20 percent of players have a significant difference in their singles and doubles ability, either because of age or hand-eye coordination,” he explained. “It would be great to see that ratings change happen in the coming years.”​

This is certainly something I think makes sense, I've written about it before (http://computerratings.blogspot.com/2014/03/should-there-be-separate-ntrp-ratings.html and http://computerratings.blogspot.com/2015/03/is-this-usta-league-player-tanking-to.html), but I had not seen anything from the USTA hinting that they agreed or would be open to doing it.

Now, it introduces a lot of complications as what do you do if a player has different singles and doubles ratings? What teams can they sign-up for and presumably they become limited to just playing singles or doubles on different teams?

So it is interesting to think about, but I'd still be surprised to see the USTA do it.

LOL at this. USTA can't even keep good fair ratings for each division.. how are they going to keep accurate rating for singles and doubles separately. their computer program isn't capable of it. Unless they paid Google to write a entirely new Algorithm,,,,,,,,,,,,,,,,lol

 

[wings56]

They have a mixed exclusive rating. ..I can see it happening

 

[g4driver]

If the USTA listened to Schmke's recommendations or paid him to write and update the algorithm, I would have more faith in the USTA

Had to edit the first word "It" to "If"- sorry typos via my iPhone and Tapatalk.

 

[Cindysphinx]

i would love that. i have no business on a 4.0 singles court. but if i could join a 3.5 team and play singles, i might consider it.

for womens teams, it would make it much easier to find singles players for 3.5 and 3.0.

the age division singles for 40 plus is still way too demanding.

 

[PBODY99]

There really isn't any reason not to rate players singles vs doubles. A;; of the data is thee.

 

[Max G.]

That would be pretty cool!

I suspect that for most players, their S/D rankings would be relatively close, but for some people on the margin one would be 0.5 lower or higher than the other (especially people who don't move very well, which might hurt them in singles far more than in doubles).

 

[Moveforwardalways]

How can you possibly have an accurate doubles rating when all of your match outcomes are determined/largely influenced by how your partner plays? No way to be accurate.

 

[schmke]

How can you possibly have an accurate doubles rating when all of your match outcomes are determined/largely influenced by how your partner plays? No way to be accurate.

Because you factor in the strength of your doubles partners. It isn't really that hard, and only can have issues if two players play together all the time as then, it can't accurate reflect the changes due to one player getting better/worse vs the other. But if a player plays with a few different partners over the year, the algorithm works pretty well.

 

[Moveforwardalways]

Because you factor in the strength of your doubles partners. It isn't really that hard, and only can have issues if two players play together all the time as then, it can't accurate reflect the changes due to one player getting better/worse vs the other. But if a player plays with a few different partners over the year, the algorithm works pretty well.

Actually, it is that hard. No way the USTA algorithm is statistically valid for doubles. NTRP wasn't even designed to measure doubles skills. It's a farce.

 

[Nostradamus]

So what happens if you play Singles only once or twice in a given year ? Does that give you accurate singles rating....? I think Not

 

[Max G.]

Actually, it is that hard. No way the USTA algorithm is statistically valid for doubles. NTRP wasn't even designed to measure doubles skills. It's a farce.

The text of the descriptions of ntrp levels doesn't really affect the rating algorithm, though. It's just a way to help people find the right level initially.

I mean, fundamentally, the algorithm is pretty simple. It adds up the ntrp ratings of you and your partner, compares to the added ntrp ratings of your opponents. From this comparison it comes up with how many games it "expected" each team to win. If you and your partner did better than your ratings would indicate, your rating goes up. If you did worse than the ratings would indicate, your ratings go down. There's a little bit of math involved in calculating the expected difference in games won, and in updating the ratings, but this is all pretty standard stuff, it's not like the USTA needs to make up its own new algorithm for this.

It works fine for both singles and doubles. The main caveat is that when you play doubles and do better or worse than expected, the algorithm doesn't know which one of you was responsible, so both players get bumps; but as schemke said, if you play with a bunch of different partners, it'll work out on average.

 

[J_R_B]

This would be awesome if there were unlimited matches for every player, but part a big of the issue with the ratings is statistical credibility. As it currently stands, there are a lot of players that they try to rate with just 2 or 3 matches, and the rating just isn't all that accurate with limited experience. Now, you need to play 3 singles AND 3 doubles matches to get an accurate rating? You're going to end up with a bunch more people that you're rating with a very small sample size, leading to ratings with a much higher variance. For the people that play at least 5 or 6 matches of both singles and doubles, this is a great idea. Unfortunately, that's just not that many players.

 

[schmke]

This would be awesome if there were unlimited matches for every player, but part a big of the issue with the ratings is statistical credibility. As it currently stands, there are a lot of players that they try to rate with just 2 or 3 matches, and the rating just isn't all that accurate with limited experience. Now, you need to play 3 singles AND 3 doubles matches to get an accurate rating? You're going to end up with a bunch more people that you're rating with a very small sample size, leading to ratings with a much higher variance. For the people that play at least 5 or 6 matches of both singles and doubles, this is a great idea. Unfortunately, that's just not that many players.

Excellent point, and part of the reason I say I think it is unlikely the USTA does it.

FWIW, it appears that only about 10% of the players that play 18+ and 40+ play 3 or more singles and doubles matches, so this would seemingly apply to only those players. And if you raise the bar to 5 to try and have more accurate ratings as J_R_B notes, the number drops to 4%.

 

[tennixpl]

that's a whole lotta effort for such a small percentage of players. As a player of both, eh! don't really consider it an issue. What problem is this trying to fix or make better. a really good singles player who sucks at doubles...well he just wont play much doubles anyway. So a 4.5 singles guy want to hang out at 3.5 doubles??? A complicated solution in search of a problem.

 

[onehandbh]

Rumor has it that they will be adding three additional qualifiers to the ratings, which currenlty have B=benchmark, S=self rated, C=computer rated.

e.g.
4.0F = Falling (falling 4.0 is someone that is a 4.0 but getting worse due to age, injury, incorporating too many tips from random TW posters, etc.)
4.0R = Rising (a 4.0 that is currenlty improving)
4.0P = Peak (the player's peak rating when they are playing really well and zoning. They could normally be a 2.5, but every once in awhile they crush a forehand off of the frame and it takes a weird bounce that can trouble 4.0 players. This is also the same as their TW self rating).

 

[cknobman]

Actually, it is that hard. No way the USTA algorithm is statistically valid for doubles. NTRP wasn't even designed to measure doubles skills. It's a farce.

I'd say its you, and not the USTA, that has the problem if you choose to repeatedly and elusively play with a crappy partner that is causing you to lose all of your matches.

Either that or it is you that actually sucks.

 

[g4driver]

As I guy who plays both singles and doubles, I actually prefer the USTA stay with one rating.

But there wouldn't be any requirement for a self-rated player to play three matches of each if they followed my suggestion, below.

The only issue I see is the self-rated players getting an initial rating. There would be no issue if the player were required to play three singles OR three doubles matches to establish the initial rating.

If the USTA went with the dual rating system,
have the USTA make a rule that the self-rated player must play three singles matches OR three doubles matches, before they can be used in the other role.

As a Captain, I would just follow the USTA mandate. That wouldn't be hard for me or any Captain to follow that rule.

E.g., If a self-rated player was required to play either three singles matches OR three doubles marches, before playing the other, those three matches would establish a DNTRP for the player.

Let's say the player is put in lineups at doubles. Once the player had played three doubles matches in the spring leagues, they could then play singles if needed. If the USTA gave the player one rating at this point it could serve as his rating for both just like the computer rated players who only play one or the other. And like any computer rated player, only when that player had played three more singles matches, would they generate the second rating.

If the doubles and singles ratings were the same in the USTA computer, the player would only have one number. Only those with two different ratings should have the two numbers by their name. This isn't as complicated as it appears IMO, but honestly, I hope this system never happens. I think the current system works just fine .