PDA

View Full Version : USTA should use a rating system similar to chess/table tennis


lethalphorce
12-01-2009, 01:02 PM
I think the USTA should take a good look at the ELO type rating system that chess, table tennis & a number of other competitive activities use. It seems much more accurate.

For those of you that aren't familiar with it. .. basically everyone has a numerical rating. When you play someone else, your rating is weighed against your opponent's and if you win, your rating goes up. . . lose and it goes down. The disparity between you & your opponent's ratings will determine how much your rating will change. The bigger the upset, the bigger the gain for the underdog and the bigger the loss for the favorite.

When I played in table tennis leagues years ago, everything was pretty clearcut. Everyone had a much better idea of where they stood.

kylebarendrick
12-01-2009, 01:20 PM
I am pretty sure the NTRP system is already similar to that. Everyone has a numerical rating. The computer predicts a margin of victory based on the differential between your rating and your opponent's. If you win by more than that, then your rating goes up. If you win by less (or lose) then your rating goes down.

They just don't make your exact rating available to discourage people from throwing games/sets/matches.

KFwinds
12-01-2009, 03:06 PM
I think the USTA should take a good look at the ELO type rating system that chess, table tennis & a number of other competitive activities use. It seems much more accurate.

For those of you that aren't familiar with it. .. basically everyone has a numerical rating. When you play someone else, your rating is weighed against your opponent's and if you win, your rating goes up. . . lose and it goes down. The disparity between you & your opponent's ratings will determine how much your rating will change. The bigger the upset, the bigger the gain for the underdog and the bigger the loss for the favorite.

When I played in table tennis leagues years ago, everything was pretty clearcut. Everyone had a much better idea of where they stood.

Umm...that's pretty much the way the USTA currently calculates ratings.

Jim A
12-01-2009, 03:17 PM
someone loses every match, just because there's a loser doesn't mean the system is wrong

even now there is for many of us quite a few months before league starts...my goal is to be a competitive 3.5 6 months from now and go from there...if its enough to get me bumped to 4.0, great..if not ..try again

lethalphorce
12-01-2009, 03:34 PM
someone loses every match, just because there's a loser doesn't mean the system is wrong


I'm not saying the current system is wrong. . . I'm saying that I like the other system better. It was just more precise. Instead of having 95% of players bunched into 4 groups (3.0 - 4.5), you'd have a much more linear distribution.

Swissv2
12-01-2009, 03:40 PM
I wonder something similar has been attempted before?

Jim A
12-01-2009, 03:48 PM
A computer can't tell you the outcome of a match or predict it, but its the system we have in place. Would a computer have predicted Soderling beat Nadal on clay in Paris, or anyone? Giants over Patriots, etc

So it provides a general range of players, some will be better others even and more below what the computer believes to happen. It has to be large enough to ensure enough players take part and support it, while keeping a general level of competition.

When I look at the USTA leagues in my area (Intermountain) for the most part there are 1-3 teams out of 8-9 battling for the top spot, a slough of teams right around .500 and a couple at the bottom who still win a couple matches along the way and occasionally pull the big upset.

For the most part, they play for the camaraderie and very few are built for sectional/national competition. There will always be players who are at a level above their rating, what I'd like to see is the USTA be more pro-active at the local level at weeding out egregious sandbaggers who have either taken 3 years off, lied on the self-rate form, etc with significant penalties for the player, captain and to a lesser extent other teammates (i.e lifetime for player/captain, 1 year for all team members...problem solved..and a lot fewer headaches from 8 *****ing teams out of 9 regarding a handful of players)

Cruzer
12-01-2009, 03:59 PM
I think the USTA should take a good look at the ELO type rating system that chess, table tennis & a number of other competitive activities use. It seems much more accurate.

For those of you that aren't familiar with it. .. basically everyone has a numerical rating. When you play someone else, your rating is weighed against your opponent's and if you win, your rating goes up. . . lose and it goes down. The disparity between you & your opponent's ratings will determine how much your rating will change. The bigger the upset, the bigger the gain for the underdog and the bigger the loss for the favorite.

When I played in table tennis leagues years ago, everything was pretty clearcut. Everyone had a much better idea of where they stood.

What you describe appears similar to the dynamic rating system that was played around with a few years ago. Basically your rating could change every week depending on your match results. Norcal applied it to a few leagues but it never had a lot of support and it quietly went away.

OrangePower
12-01-2009, 04:12 PM
I'm not saying the current system is wrong. . . I'm saying that I like the other system better. It was just more precise. Instead of having 95% of players bunched into 4 groups (3.0 - 4.5), you'd have a much more linear distribution.

As others have noted, rating are actually recalculated dynamically after every match you play, and again recalculated at the end of the season... and the distribution is much more granular (precise to 0.01). However, the USTA chooses not to make these exact ratings public - for public consumption they round up to the nearest 0.5.

What you describe appears similar to the dynamic rating system that was played around with a few years ago. Basically your rating could change every week depending on your match results. Norcal applied it to a few leagues but it never had a lot of support and it quietly went away.

Dynamic rating is still calculated, and is still used for determining things such as DQs during the course of a season. But when made public it just makes it so much easier for individuals to manipulate their ratings.

JavierLW
12-01-2009, 08:33 PM
I think the USTA should take a good look at the ELO type rating system that chess, table tennis & a number of other competitive activities use. It seems much more accurate.

For those of you that aren't familiar with it. .. basically everyone has a numerical rating. When you play someone else, your rating is weighed against your opponent's and if you win, your rating goes up. . . lose and it goes down. The disparity between you & your opponent's ratings will determine how much your rating will change. The bigger the upset, the bigger the gain for the underdog and the bigger the loss for the favorite.

When I played in table tennis leagues years ago, everything was pretty clearcut. Everyone had a much better idea of where they stood.

Yes, the problem is the NTRP system was designed to be a sort of description of skill, not as a competitive separator just for leagues.

But obviously tennis is a sport so other then the whole teaching and working on your game process, the practical application is to divide leagues in divisions based on skill which is never going to be perfect.

But actually if you knew more about how tennis ratings worked, they actually already do something similar to what you are describing. Ratings are based on the difference in games (which is more precise then just wins or losses or draws), versus your opponents actual exact rating. It's really the same thing, just more is hidden....

JavierLW
12-01-2009, 08:38 PM
I'm not saying the current system is wrong. . . I'm saying that I like the other system better. It was just more precise. Instead of having 95% of players bunched into 4 groups (3.0 - 4.5), you'd have a much more linear distribution.

I used to play chess so Im familar with chess ratings.

It is NOT linear. If you get 100 people in a room and teach them chess and have them go play a bunch of tournaments most will end up below 1000 or probably nowhere past 1500.

But the difference is in chess you dont have special chess leagues for players under 1000, 1000-1200, 1200-1400, etc..... You usually play tournaments and you can pretty much play anyone regardless of their rating.

If anything Chess ratings are more like tennis ratings actually except that they didnt necessarily try to describe what having a 2200+ rating means, it's just more competition (wins and losses) based since it's just a simple game so the translation from these ratings to "who's got the mad chess skills" is more accurate.

If people really want a linear system they will cry about chess as well. They'll say WAY more people should be a the highest level just like the lowest level, so I guess more people can say they are grandmasters then.....

lethalphorce
12-02-2009, 04:54 AM
But actually if you knew more about how tennis ratings worked, they actually already do something similar to what you are describing. Ratings are based on the difference in games (which is more precise then just wins or losses or draws), versus your opponents actual exact rating. It's really the same thing, just more is hidden....

Ah, I see. I didn't know they worked like that already. I wish it weren't hidden.

Part of what I liked about playing in my table tennis league was that your rating was a pretty accurate measure of how good a player you were. . . and everyone's ratings were there for all to see. As a competitive person (which I'm assuming most of us are), you CARED about your rating. You wanted to increase it because essentially, increasing your rating meant you were getting better. And the rating changes were immediate. On an average night we would play 4-5 matches. . . the next day, we would have a new rating. You didn't have to wait a year.

The NTRP just seems. . . murkier.

raiden031
12-02-2009, 05:48 AM
Ah, I see. I didn't know they worked like that already. I wish it weren't hidden.

Part of what I liked about playing in my table tennis league was that your rating was a pretty accurate measure of how good a player you were. . . and everyone's ratings were there for all to see. As a competitive person (which I'm assuming most of us are), you CARED about your rating. You wanted to increase it because essentially, increasing your rating meant you were getting better. And the rating changes were immediate. On an average night we would play 4-5 matches. . . the next day, we would have a new rating. You didn't have to wait a year.

The NTRP just seems. . . murkier.

I think that the NTRP rating calculation is very good for accurately representing the true skill level of a player.

The problem is not the rating calcuation, it is the the NTRP league format which consists of teams and multiple skill divisions in which a winning team can advance to National championships at any level.

Chess is different because it is for the most part individual. You can get online and play chess against anybody (there are no arbitrary boundaries that you are bound to), yet you more than likely seek out the highest rated opponents who you can reasonably compete against. In chess, there is no incentive to keep your rating low like in tennis, where you keep your rating artificially low in order to play on a team in a lower skill division to advance to Nationals.

Plus in chess there is not as much administration needed to run competitions, especially with all the online stuff available. In tennis you are at the mercy of whatever competitive programs are available. Many areas don't have enough high level players so the high level players are forced to play at lower levels in order to find any competition at all.

UnforcedError
12-02-2009, 08:09 AM
I think that the NTRP rating calculation is very good for accurately representing the true skill level of a player.

The problem is not the rating calcuation, it is the the NTRP league format which consists of teams and multiple skill divisions in which a winning team can advance to National championships at any level.

Chess is different because it is for the most part individual. You can get online and play chess against anybody (there are no arbitrary boundaries that you are bound to), yet you more than likely seek out the highest rated opponents who you can reasonably compete against. In chess, there is no incentive to keep your rating low like in tennis, where you keep your rating artificially low in order to play on a team in a lower skill division to advance to Nationals.

Plus in chess there is not as much administration needed to run competitions, especially with all the online stuff available. In tennis you are at the mercy of whatever competitive programs are available. Many areas don't have enough high level players so the high level players are forced to play at lower levels in order to find any competition at all.

Actually in chess if you play tournaments on or offline there are categories divided by ELO rating. What always seemed really odd to me about many chess tournaments is that they even had prize money at lower levels and I always thought it was strange that the best player at some poor level can make money, it certainly encouraged sandbagging to that level. If you are just playing for fun then yes often you seek out the best players you can, just like tennis.

I always thought the whole sectionals/nationals thing about USTA was odd too, after all what does it really mean if you get to nationals of some mediocre division? Probably means your team has a high number of people that belong at another level, congratulations.

JavierLW
12-02-2009, 08:15 AM
Ah, I see. I didn't know they worked like that already. I wish it weren't hidden.

Part of what I liked about playing in my table tennis league was that your rating was a pretty accurate measure of how good a player you were. . . and everyone's ratings were there for all to see. As a competitive person (which I'm assuming most of us are), you CARED about your rating. You wanted to increase it because essentially, increasing your rating meant you were getting better. And the rating changes were immediate. On an average night we would play 4-5 matches. . . the next day, we would have a new rating. You didn't have to wait a year.

The NTRP just seems. . . murkier.

I havent played much table tennis. I know they went from 11 point games to 7 point games or something though didnt they? How many of those "games" do you play in a particular match?

Because I think the big difference is you can play many matches in a small period of time, and what constitutes a "game" (the lowest measurable scoring method unless you want to get into points) consists of more of the match then it does in tennis. (a tennis match consists of many games)

I think that could be the big difference.

If you want to rate SKILL and not just simple wins and losses, you go to the lowest easiest measurable unit.

In tennis there is a certain amount of skill involved to win one single game and apparently someone felt it was useful to measure that.

You get a better sense of whether the match was "competitive" or not that way. (and even that's not perfect you can have "competitive" games or short blow out games but that's hard to measure)

A typical tennis match can take from 90 minutes to 2 or 3 hours sometimes so it's probably worth it to care about more then just who won or who lost.

So in tennis if you measure it the way they do it will sometimes end up pretty far off from who actually won the match.

In table tennis it's probably a lot closer so people who just care about simple wins and losses will think it's better.

In chess, that's all you have is wins, losses, draws. There is no smaller measure of skill when you're talking about one chess match. So the skill rating is seemingly directly associated with your record.

JavierLW
12-02-2009, 08:24 AM
I always thought the whole sectionals/nationals thing about USTA was odd too, after all what does it really mean if you get to nationals of some mediocre division? Probably means your team has a high number of people that belong at another level, congratulations.

I feel the same way, although it's probably a fun experience.

This is besides the point, but Im in a Renegade League with no rating system at all, you just pick a rating that you feel like playing in and the only real drawback is if you sandbag too much you will get "frowned upon" and if it's really bad the league coordinator might "have a talking too with you".

It has a 3.5 and a 4.0 and a 4.5 level. This league has been around since around 2000 (they all were USTA leagues that left the league and formed their own teams).

It's common that you may find a 4.0 guy in your 3.5 league, sometimes he's even 4.0 in USTA.

But because there are no nationals to get too (we just have local playoffs were the first place team all receives a hat), you won find anything to the extent that's been in USTA. (you dont need to find 4.5 and 5.0 players for your "3.5 SUPERTEAM")

And Im not saying that all teams that go to nationals are like that, but there are a lot of teams where that's their goal and they will do that. Especially if they've clobbered everyone in their local league one year and made it to sectionals and found they get get past that.

That's part of the hard part of having a competitive league like this where we're divided into skill ratings.

If you treat it like a league where you're trying to get ahead then you'll naturally find players that can "compete" better with the other teams players. The more levels of playoffs you have to go thru or want to get too, the more screwed up things get.

I know for my team last year I would of been happy if we won our State (which we were 4 points away from even getting to that playoff). It used to be called "State League" here for a reason.

Beyond going to sectionals though I dont know how much I really cared. They penalize you $500 for one if you go to Nationals, and I agree with you, at some point I wonder how much of a big deal is it (to justify the cost of paying for our travel, hotels, etc.....)? I could brag that we're state champions and that sounds cool to all of my non-tennis playing friends, but do they really care where I went beyond that? I doubt it....

Cruzer
12-02-2009, 09:36 PM
Dynamic rating is still calculated, and is still used for determining things such as DQs during the course of a season. But when made public it just makes it so much easier for individuals to manipulate their ratings.

Well of course dynamic ratings are still calculated except the USTA doesn't publish them anymore. Very few players manipulated their ratings when dynamic ratings were published since there was no point in doing so however there were other issues that made using dynamic ratings unwieldy.

dlk
12-03-2009, 03:29 AM
Let's do the BCS system. We can't go wrong:-| No really, don't know if financially feasible, but a comuter system mixed w/club pro analysis & tournament result analysis.