Alex Table Tennis - MyTableTennis.NET Homepage
  Help Desk Help Desk  New Posts New Posts RSS Feed - USATT ratings calculator algorithm..??
  FAQ FAQ  Forum Search   Events   Register Register  Login Login

USATT ratings calculator algorithm..??

 Post Reply Post Reply
Author
angelleye View Drop Down
Member
Member
Avatar

Joined: 08/05/2014
Location: Kansas City
Status: Offline
Points: 48
Post Options Post Options   Thanks (0) Thanks(0)   Quote angelleye Quote  Post ReplyReply Direct Link To This Post Topic: USATT ratings calculator algorithm..??
    Posted: 05/05/2021 at 1:18pm
I am looking to create my own ratings calculator to use for our local club's non-sanctioned leagues / tournaments.


That's helpful, but I'm wondering if anybody knows the actual algorithm I could use to plug in our data sets and have it spit out the final rating results..??

Any information on this would be greatly appreciated. Thanks!
Back to Top
Sponsored Links


Back to Top
Egghead View Drop Down
Premier Member
Premier Member
Avatar

Joined: 09/05/2009
Location: N.A.
Status: Offline
Points: 4115
Post Options Post Options   Thanks (0) Thanks(0)   Quote Egghead Quote  Post ReplyReply Direct Link To This Post Posted: 05/06/2021 at 2:03pm
Originally posted by angelleye angelleye wrote:

I am looking to create my own ratings calculator to use for our local club's non-sanctioned leagues / tournaments.


That's helpful, but I'm wondering if anybody knows the actual algorithm I could use to plug in our data sets and have it spit out the final rating results..??

Any information on this would be greatly appreciated. Thanks!
Few years ago, I wrote a small app to calculate rating. I find that the problem with US rating system is their +50 (??) rating no deduction rule. Some players like it but some players don't.
Aurora ST: Rhyzm / Talent OX
Back to Top
David Marcus View Drop Down
Super Member
Super Member
Avatar

Joined: 10/28/2010
Location: Somerville, MA
Status: Offline
Points: 129
Post Options Post Options   Thanks (0) Thanks(0)   Quote David Marcus Quote  Post ReplyReply Direct Link To This Post Posted: 05/09/2021 at 9:53am
I only skimmed the USATT webpage, but it looks like it does describe the algorithm. Which part is not clear?

If you want more accurate, easy-to-use ratings, there is https://www.ratingscentral.com/. It isn't free, but the cost is reasonable.

Back to Top
dyger View Drop Down
Beginner
Beginner


Joined: 05/10/2021
Location: From UK in US
Status: Offline
Points: 7
Post Options Post Options   Thanks (0) Thanks(0)   Quote dyger Quote  Post ReplyReply Direct Link To This Post Posted: 05/11/2021 at 2:12pm
Originally posted by David Marcus David Marcus wrote:

If you want more accurate, easy-to-use ratings, there is https://www.ratingscentral.com/. It isn't free, but the cost is reasonable.

Hello sir,
I am not too familiar but I was wondering why you say ratingCentral is more accurate & easy-to-use

Also if there is a cost anyway, am I not better off with a USATT sanctioned tournaments since they provide other services to members.

Also, I may be wrong but aren't your league ratings mixed up with tournament ratings ?

Finally I see your ratings go over 3000 while it seemed USATT keeps them below 3000. Are there some reasons for either approach ?

 
Back to Top
David Marcus View Drop Down
Super Member
Super Member
Avatar

Joined: 10/28/2010
Location: Somerville, MA
Status: Offline
Points: 129
Post Options Post Options   Thanks (0) Thanks(0)   Quote David Marcus Quote  Post ReplyReply Direct Link To This Post Posted: 05/11/2021 at 5:06pm
> I am not too familiar but I was wondering why you say ratingCentral is
> more accurate & easy-to-use

The Ratings Central algorithm was originally developed when I was on the USATT Ratings Committee to fix the problems with the USATT rating system. We did a careful analysis that compared the two rating systems. The Ratings Committee unanimously recommended that USATT adopt the new system. But by then the USATT board was dominated by clueless people.

As for easy to use, there are multiple easy ways to submit data, and the website is easy for players to use to see their results and ratings. Ratings Central will optionally email a player to let them know that a new event that they played in has been posted on the website.

> Also if there is a cost anyway, am I not better off with a USATT
> sanctioned tournaments since they provide other services to members.

That's for you to decide. But, the cost for Ratings Central is reasonable for what you get.

> Also, I may be wrong but aren't your league ratings mixed up with
> tournament ratings ?

That is a feature, not a bug. You can submit whatever tournaments or leagues that your club runs.

> Finally I see your ratings go over 3000 while it seemed USATT keeps them
> below 3000. Are there some reasons for either approach ?

The best players in Ratings Central (e.g., Fan Zhendong, Ma Long) are much better than the best players in the USATT system. But, the USATT rating scale is no longer the same as the Ratings Central rating scale. They were the same in 2004 when Ratings Central was launched. Since then, USATT ratings have inflated. For example, Kanak Jha is 2746 USATT, but 2555 Ratings Central. Of course, these two ratings for Kanak Jha are based on different data. But, this does provide a rough idea of the difference in the rating scales.
Back to Top
larrytt View Drop Down
Silver Member
Silver Member
Avatar

Joined: 04/04/2005
Location: United States
Status: Offline
Points: 923
Post Options Post Options   Thanks (0) Thanks(0)   Quote larrytt Quote  Post ReplyReply Direct Link To This Post Posted: 05/12/2021 at 7:23pm

>But, the USATT rating scale is no longer the same as the Ratings Central rating scale. They were the same in 2004 when Ratings Central was launched. Since then, USATT ratings have inflated. For example, Kanak Jha is 2746 USATT, but 2555 Ratings Central. Of course, these two ratings for Kanak Jha are based on different data. But, this does provide a rough idea of the difference in the rating scales.

You wrote that the USATT ratings and Ratings Central "were the same in 2004 when Ratings Central was launched." I have the Jan/Feb 2004 USATT Magazine in front of me, with the ratings. In "Top Olympic Eligible Men," the #1 player is David Zhuang at 2726. According to the ITTF rankings, he was #176 in the world at the time. ITTF only seems to have rankings going back to 2001, and Zhuang's highest in that time was #105, with him on the US National Team. From this, and similar players rated in the 2700 range, it's obvious that a USATT 2700 rating generally doesn’t put you in the top 100 in the world, at least not since 2001 or so. Kanak is currently #30 in the world (and #33 when he got the 2746 rating before the pandemic, was as high as #22) and has won four consecutive men's singles titles at the Nationals, and should be considerably higher than Zhuang's 2726. So if anything, Kanak's current USATT rating of 2746 is too low, as opposed to being inflated.

Ratings Central has Kanak at 2555. In 2004, a 2555 rating would make Kanak #10 in the US, behind David Zhuang, Cheng Yinghua, Barney J. Reed, Mark Hazinski, Eric Owens, Jim Butler, Todd Sweeris, Khoa Nguyen, and Darko Rop. Unless someone wants to make the argument that all of these players were as good or better than Kanak is now, and Zhuang about 150 points better, I think we can agree that since 2004, at least at the US higher levels, the ratings in Ratings Central have deflated about 200 or more points, while USATT ratings have been relatively stable. (USATT ratings had a period of inflation, but that was farther back.) 

For comparison, Cheng Yinghua had numerous ratings around 2800 for much of the 1990s, with world rankings often in the top 30 or 40 or so. Jim Butler had ratings often in the low-to-mid 2700s, and often was ranked in the 70-100 range. So it seems reasonable that Kanak should be rated close to 2800. So his 2746 rating seems reasonable, just a bit too low.

-Larry Hodges



Edited by larrytt - 05/12/2021 at 8:04pm
Professional Table Tennis Coach & Writer
Member, USATT Hall of Fame
USATT National & ITTF Certified Coach
Former Chair, USATT Coaching Committee
www.TableTennisCoaching.com
www.MDTTC.com
Back to Top
David Marcus View Drop Down
Super Member
Super Member
Avatar

Joined: 10/28/2010
Location: Somerville, MA
Status: Offline
Points: 129
Post Options Post Options   Thanks (0) Thanks(0)   Quote David Marcus Quote  Post ReplyReply Direct Link To This Post Posted: 05/12/2021 at 7:47pm
Unfortunately, you can't do comparisons of rating systems by looking at a few high-rated players using three rating/ranking systems of varying accuracies.
Back to Top
stiltt View Drop Down
Assistant Admin
Assistant Admin
Avatar

Joined: 07/15/2007
Location: USA
Status: Offline
Points: 682
Post Options Post Options   Thanks (0) Thanks(0)   Quote stiltt Quote  Post ReplyReply Direct Link To This Post Posted: 05/12/2021 at 8:00pm
From the little I understand, the top winning players in the USA can't see their ratings go up much because they hit a ceiling when there are not many points they can grab from other American players. The ratings inflation concerns more the crowd than the cream. Is this correct?
Back to Top
larrytt View Drop Down
Silver Member
Silver Member
Avatar

Joined: 04/04/2005
Location: United States
Status: Offline
Points: 923
Post Options Post Options   Thanks (0) Thanks(0)   Quote larrytt Quote  Post ReplyReply Direct Link To This Post Posted: 05/12/2021 at 8:03pm
Originally posted by David Marcus David Marcus wrote:

Unfortunately, you can't do comparisons of rating systems by looking at a few high-rated players using three rating/ranking systems of varying accuracies.

You can make certain comparisons, as I did. If Ratings Central hasn't deflated (or inflated) since 2004, then Kanak's current 2555 in Ratings Central would be equally accurate in 2004, which would mean he would only be #10 back then, and 150 points weaker than David Zhuang. Since this is obviously not true, we can conclude, as I wrote, that somehow, his rating is much too low in Ratings Central. If you want more data, Nikhil Kumar is actually the US #1 at 2767. In Ratings Central he is 2440. In the 2004 ratings, the rankings only go to #20, which is Ashu Jain at 2447, so Nikhil wouldn't even make the top 20. (Nikhil played in 12 events processed by RC in 2019-2020.) So obviously the ratings in Rating Central for Kanak, Nikhil, and the other top US players I looked are way too low (relative to both current and 2004 USATT ratings), and the RC ratings, at least for higher US players, have deflated quite a bit. 
-Larry Hodges
Professional Table Tennis Coach & Writer
Member, USATT Hall of Fame
USATT National & ITTF Certified Coach
Former Chair, USATT Coaching Committee
www.TableTennisCoaching.com
www.MDTTC.com
Back to Top
David Marcus View Drop Down
Super Member
Super Member
Avatar

Joined: 10/28/2010
Location: Somerville, MA
Status: Offline
Points: 129
Post Options Post Options   Thanks (0) Thanks(0)   Quote David Marcus Quote  Post ReplyReply Direct Link To This Post Posted: 05/12/2021 at 9:03pm
Originally posted by stiltt stiltt wrote:

From the little I understand, the top winning players in the USA can't see their ratings go up much because they hit a ceiling when there are not many points they can grab from other American players. The ratings inflation concerns more the crowd than the cream. Is this correct?
Inflation is, by definition, a general increase in ratings (compared to the playing strength). The effect on particular players may vary.

The USATT rating system can have problems with the top players. But, it has many problems. The system treats ratings based on a few matches the same as ratings based on many matches, which is illogical and causes experienced players to feel they need to protect their ratings since one loss can lose them a lot of points. The system can have problems keeping up with rapidly improving players. The ratings for players who are playing consistently tend to bounce around too much. The easiest way to gain points is to go into a large tournament underrated and get adjusted, which encourages players to try to game the system and is unfair. These problems have been known for a long time. And, we developed a system without these problems. But, accurate ratings was not of interest for the USATT board, and apparently is still not of interest.

Back to Top
larrytt View Drop Down
Silver Member
Silver Member
Avatar

Joined: 04/04/2005
Location: United States
Status: Offline
Points: 923
Post Options Post Options   Thanks (0) Thanks(0)   Quote larrytt Quote  Post ReplyReply Direct Link To This Post Posted: 05/12/2021 at 9:26pm
Originally posted by David Marcus David Marcus wrote:

Inflation is, by definition, a general increase in ratings (compared to the playing strength). The effect on particular players may vary.
That is correct. And deflation is, by definition, a general decrease in ratings (compared to the playing strength). Using your example of Kanak Jha, with his RC rating of 2555 (and a low SD of 60), and comparing it to players from 2004 (when you said the two systems were "the same"), we can conclude there has been severe deflation in the Ratings Central ratings, at least at the higher levels. A browsing of the top players in RC shows the top few may actually have inflated rating, but once you get past the top few, the ratings deflate quite a bit, compared to 2004. 

Originally posted by David Marcus David Marcus wrote:

The USATT rating system can have problems with the top players. But, it has many problems. The system treats ratings based on a few matches the same as ratings based on many matches, which is illogical and causes experienced players to feel they need to protect their ratings since one loss can lose them a lot of points. The system can have problems keeping up with rapidly improving players. The ratings for players who are playing consistently tend to bounce around too much. The easiest way to gain points is to go into a large tournament underrated and get adjusted, which encourages players to try to game the system and is unfair. These problems have been known for a long time. And, we developed a system without these problems. But, accurate ratings was not of interest for the USATT board, and apparently is still not of interest.
I disagree with some of these assertions, and agree with others, and could also list problems with RC, but I'm not going to get into it. Both systems have strengths and weaknesses, and the "clueless" board you refer to went over them in great detail. However, I'm not interested in rehashing those arguments, and I'm pretty sure you don't want to either. 

When you posted above about how the USATT ratings had inflated, instead of RC's deflating, I considered jumping in to correct that. But when you called the eleven members of the USATT board at the time "clueless," you not only drew me in, but you also may have inadvertently showed readers a primary reason why your system, admirable as it is in so many ways, wasn't adopted. 
-Larry Hodges


Edited by larrytt - 05/12/2021 at 9:29pm
Professional Table Tennis Coach & Writer
Member, USATT Hall of Fame
USATT National & ITTF Certified Coach
Former Chair, USATT Coaching Committee
www.TableTennisCoaching.com
www.MDTTC.com
Back to Top
BRS View Drop Down
Gold Member
Gold Member


Joined: 05/08/2013
Location: United States
Status: Offline
Points: 1248
Post Options Post Options   Thanks (0) Thanks(0)   Quote BRS Quote  Post ReplyReply Direct Link To This Post Posted: 05/12/2021 at 11:13pm
Originally posted by angelleye angelleye wrote:

I am looking to create my own ratings calculator to use for our local club's non-sanctioned leagues / tournaments.


That's helpful, but I'm wondering if anybody knows the actual algorithm I could use to plug in our data sets and have it spit out the final rating results..??

Any information on this would be greatly appreciated. Thanks!

why not just use a ladder and do relegation and promotion?
Back to Top
denym View Drop Down
Beginner
Beginner


Joined: 05/12/2021
Location: Burnaby,BC
Status: Offline
Points: 6
Post Options Post Options   Thanks (0) Thanks(0)   Quote denym Quote  Post ReplyReply Direct Link To This Post Posted: 05/13/2021 at 12:07pm
Originally posted by David Marcus David Marcus wrote:

Unfortunately, you can't do comparisons of rating systems by looking at a few high-rated players using three rating/ranking systems of varying accuracies.

"three" ?  Is the third one by ITTF ? Or Canadian ?
Is there a description somewhere as to how ITTF method works ?


Back to Top
 Post Reply Post Reply
  Share Topic   

Forum Jump Forum Permissions View Drop Down

Forum Software by Web Wiz Forums® version 12.01
Copyright ©2001-2018 Web Wiz Ltd.

This page was generated in 0.141 seconds.

Become a Fan on Facebook Follow us on Twitter Web Wiz News
Forum Home | Go to the Forums | Forum Help | Disclaimer

MyTableTennis.NET is the trading name of Alex Table Tennis Ltd.

Copyright ©2003-2020 Alex Table Tennis Ltd. All rights reserved.