PDA

View Full Version : Playing at 4.0 with a 3.5 Partner?


memphisliza
12-14-2009, 02:45 PM
About half of our team got bumped up to 4.0 in the latest ratings sweep, so we decided to move our whole team up and play 4.0 together. (Nice group, no drama!) We were all strong 3.5's, and the ones who did not get moved up are really no weaker than the players who did. So I am wondering how the NTRP system handles teams where one player is 3.5 and one is 4.0? If I have my choice of equally strong partners and one is 4.0 and one is 3.5 (I am 4.0), how does the choice affect my computer ranking.....Does anyone know?

Geezer Guy
12-14-2009, 03:37 PM
Our 4.0 team for the past several years has had a significant number of 3.5 players. Until this year, the system seemed to do a good job of bumping up stronger players and bumping down the weaker players. Many of the 3.5 players on our 4.0 team played on a 3.5 team as well. However, last month almost all our 3.5 players were bumped up to 4.0. None of our 4.0 players were bumped up.

Topaz
12-14-2009, 04:04 PM
If the 3.5s are as strong as you say, you won't have anything to worry about. If they play up and post competitive scores, they themselves will more than likely be moved up next year.

If you play a few matches with the 3.5 partner and you guys get killed, then you may want to play with the 4.0.

JoelDali
12-14-2009, 06:10 PM
About half of our team got bumped up to 4.0 in the latest ratings sweep, so we decided to move our whole team up and play 4.0 together. (Nice group, no drama!) We were all strong 3.5's, and the ones who did not get moved up are really no weaker than the players who did. So I am wondering how the NTRP system handles teams where one player is 3.5 and one is 4.0? If I have my choice of equally strong partners and one is 4.0 and one is 3.5 (I am 4.0), how does the choice affect my computer ranking.....Does anyone know?

The only thing that matters is the win or loss.

I highly doubt the computer gives you a break for playing with inferior 3.5s that can't return well like your 4.0 friends.

kylebarendrick
12-14-2009, 10:07 PM
The computer maintains the rating differential between the two partners on a doubles team - in other words they both move up or down by the same amount based on the match results. So if your 4.0 and 3.5 are actually (based on the dynamic ratings) 0.07 apart before the match, they would be .07 apart after the match as well.

OrangePower
12-14-2009, 10:15 PM
If the 3.5s on the team are really as good as the 4.0s, then purely from the perspective of the rating algorithm it will make a significant difference who you partner with.

The algorithm takes your dynamic rating plus that of your partner's, and compares that sum to the sum of your opponents' ratings. Based on that, there is an expected score. If you do better than the expected scores, your and your partner's dynamic ratings go up, otherwise they go down. Note that the key measure is 'expected score', not 'win / lose'.

So: Playing with a 3.5 means that the score versus any given opponents that you and your partner are expected to get is LESS than the score you are expected to get if playing with a 4.0 partner. Which means that your personal rating will go up faster playing with the 3.5 partner IF you really think you can do just as well with the 3.5 versus with a 4.0.

Whether that's a good or bad thing is up to you to decide...

Topaz
12-15-2009, 04:12 AM
The only thing that matters is the win or loss.

I highly doubt the computer gives you a break for playing with inferior 3.5s that can't return well like your 4.0 friends.

Actually, no, the win/loss record is not part of the algorithm used for NTRP calculation.

JoelDali
12-15-2009, 05:52 AM
Actually, no, the win/loss record is not part of the algorithm used for NTRP calculation.

Thanks...I guess what I was trying to say is ratings are not affected by the bad weather, your partner's bad knee, or the fresh knife wound in your back.

Enjoy your time with the 3.5ers, don't stress on NTRP or tt post counts. Focus on solid volleys and Fedace whispering sweet advice in your ear and handing you a hot towel on changeovers.

:)


Sorry, I'm so hungover. Damn Hanakuh parties.

:(

http://www.conspirazzi.com/wp-content/uploads/2009/08/mel-gibson-drunk.jpg

Bud
12-21-2009, 04:16 PM
Actually, no, the win/loss record is not part of the algorithm used for NTRP calculation.

Please clarify :)

JavierLW
12-21-2009, 04:35 PM
About half of our team got bumped up to 4.0 in the latest ratings sweep, so we decided to move our whole team up and play 4.0 together. (Nice group, no drama!) We were all strong 3.5's, and the ones who did not get moved up are really no weaker than the players who did. So I am wondering how the NTRP system handles teams where one player is 3.5 and one is 4.0? If I have my choice of equally strong partners and one is 4.0 and one is 3.5 (I am 4.0), how does the choice affect my computer ranking.....Does anyone know?

I think it keeps your rating difference.

So your team generates a rating based on games won or loss against your opponent's team rating.

Then when they divide it up, they keep the same rating difference. (so the lower your partner's NTRP, the higher your rating outcome will be)

However we dont know what someone's exact dynamic rating is, so it's usually not really worth thinking about. It could be that the 3.5 in question is not much further below the 4.0's in question for all we know.

Topaz
12-21-2009, 05:35 PM
Please clarify :)

Exactly what I said.

W/L record is not used to compute you NTRP. You can go 12-0 and not get moved up if your scores do not dominate the other players at your level and/or they were close scores against people playing up. On the other hand, you can go winless or have a losing record and get moved up if you had competitive scores against strong players at your level.

The computer uses your dynamic NTRP, that of your opponent(s), (also that of your partner if you're playing dubs), predicts an outcome, and then, when scores are entered, adjusts everyone's dynamic NTRP based on the comparison of the prediction to the actual results.

This site explains it well:

http://besttennislinks.com/USTA%20Ratings%20and%20How%20the%20System%20Works. htm


From the site:

One thing that most players don't know is that the USTA rating system doesn't care if you win or lose. The system only wants to know how well you performed so it looks at your game win/lost ratio. You can lose a match and have your rating go up by winning more games
than you lost. Example: you lose the first set in a tiebreaker 6-7, then you win the second set 6-0, and loose the 3rd set in another 6-7 tiebreaker. You won 18 games and your opponents only won 14 games. Your team captain isn't happy, but your rating goes up! A lost match in this case is really a big win as far as your rating goes.

JoelDali
12-21-2009, 06:16 PM
With all due respect, who is Richard Goulet and has he seen the latest source code of TennisLink?

Until then, its all speculation even if is based on heresay or common sense, right?

I'm sure there are instances of people that were bumped or not bumped that don't fit into Mr. Goulet's presented criteria.

Just playing devil's advocate here. I'm bored, sorry...lol.

Maybe Richard = Robert?

http://www.epgold.com/news/pics-october-2007/Robert_Goulet.sff_NYRG106_20071030193719.jpg

Topaz
12-21-2009, 06:45 PM
Actually, no, it is not heresay...those of us who have played long enough know how it works and/or have had the chance to speak to USTA rating officials about how it works. You can choose not to believe it, but that won't change how it works.

Ripper014
12-23-2009, 01:53 PM
Exactly what I said.

W/L record is not used to compute you NTRP. You can go 12-0 and not get moved up if your scores do not dominate the other players at your level and/or they were close scores against people playing up. On the other hand, you can go winless or have a losing record and get moved up if you had competitive scores against strong players at your level.

The computer uses your dynamic NTRP, that of your opponent(s), (also that of your partner if you're playing dubs), predicts an outcome, and then, when scores are entered, adjusts everyone's dynamic NTRP based on the comparison of the prediction to the actual results.

This site explains it well:

http://besttennislinks.com/USTA%20Ratings%20and%20How%20the%20System%20Works. htm


From the site:


Yes but in general... if you win you will have more games won than your opponent... it is only the oddest of situations that you would have less games than your opponent and register a win.

Cindysphinx
12-23-2009, 02:28 PM
Yes but in general... if you win you will have more games won than your opponent... it is only the oddest of situations that you would have less games than your opponent and register a win.

All it takes to win more games and still lose is a bad set: 1-6, 6-4, 0-1.

Cindy -- who has had her share of really bad sets

Topaz
12-23-2009, 06:23 PM
Yes but in general... if you win you will have more games won than your opponent... it is only the oddest of situations that you would have less games than your opponent and register a win.

Again, the 'win' is not what I'm talking about...you can lose a match 4-6, 4-6, but if the computer predicted (based on the dynamic NTRPs of you and your partner before the match) that you were going to lose 1-6, 1-6, your dynamic NTRP will go up, even though you lost the match. NTRP is not calculated using a win/loss record. That is a huge misunderstanding here on this board.

Ripper014
12-23-2009, 06:40 PM
Again, the 'win' is not what I'm talking about...you can lose a match 4-6, 4-6, but if the computer predicted (based on the dynamic NTRPs of you and your partner before the match) that you were going to lose 1-6, 1-6, your dynamic NTRP will go up, even though you lost the match. NTRP is not calculated using a win/loss record. That is a huge misunderstanding here on this board.

Absolutely... but the computer makes its prediction based on your playing history and those of your competitors. So for you to move up... you must have some success... or those you compete with are winning... and you are competing with them. You are not going to move up being competitive with losers.

OrangePower
12-23-2009, 06:53 PM
Absolutely... but the computer makes its prediction based on your playing history and those of your competitors. So for you to move up... you must have some success... or those you compete with are winning... and you are competing with them. You are not going to move up being competitive with losers.

It depends. If you are a 3.5 and are competitive with 4.5 losers... yup, you're going to move up. It's all about your dynamic rating compared to theirs, and the actual score versus the score predicted by the rating difference.

RickGoulet
01-04-2010, 04:40 PM
With all due respect, who is Richard Goulet and has he seen the latest source code of TennisLink?

Until then, its all speculation even if is based on heresay or common sense, right?


No I have not seen the code for the algorithm that the USTA uses, but they don't try and keep how it works a secret.

A year ago I was a team captain with a lot of new players and I was having trouble figuring out where to place people in the lineup. Watching them play to estimate their playing level wasn't working for me so I decided to see if I could rank my teammates based on their TennisLink results. First I researched the Internet for as much as I could find out about how the USTA rating system worked (found some good details from the USTA) and then I created my own software to attempt to do the same thing. For a team captain it's very helpful to know the playing level of your own teammates, but it's even more helpful to know the playing levels of the teams you are going to play against. The software doesn't have to be really accurate it just has to be close enough. When you don't know anything about a list of names, any information is helpful.

I like to think that my software allows me to predict how the teams will rank at the end of the season before we even play the first match. It's possible to use the ratings that my software creates to predict match results. The players with the higher ratings as assigned by my software win 80% of the time.

By programming my own software I can see first hand how hard it can be to generate accurate ratings. I'm sure that the USTA is constantly trying to improve their software just like I am and just like me they have the freedom to play around with their algorithm with one big difference. If they mess-up it affects thousands of people.


I'm sure there are instances of people that were bumped or not bumped that don't fit into Mr. Goulet's presented criteria.

If you have an example I might be able to guess at what might have happened. But ratings are part of a dynamic system so if you just have the facts at the end of the year that doesn't allow you to know what happened, say in a match mid-season. It's not just the games won and lost, but you also you need to know the dynamic ratings for all the players at the time of the match.

JoelDali
01-04-2010, 05:57 PM
Well consider me 0wned.

:)

I think some of us look a little too deep in the NTRP magical mystery formulas.

Whether you're 4.5 or a 5, who can really tell the difference.

A guy I know that took a SCal team to nationals would smoke any 5.0 on the east coast. This is a fact. But for some odd reason at the time he was a 4.5 according to the "formula" and could dominate strongly at 4.5 at the state level.

The #s are different in every region. The #s are not perfect.
The system works for the most part without much thought.

Those that play leagues and not tournaments have a MUCH different view of the #s.

I hope no offense was taken by my post many weeks ago and I apologize for spewing your name like that but at the time I just thought some people take the ntrp a little too darn seriously. I'm not the only one I'm sure.

:)