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

USATT Rating Adjustment?

 Post Reply Post Reply
Author
DonnOlsen View Drop Down
Gold Member
Gold Member
Avatar

Joined: 11/15/2008
Location: Maryland, USA
Status: Offline
Points: 1635
Post Options Post Options   Thanks (0) Thanks(0)   Quote DonnOlsen Quote  Post ReplyReply Direct Link To This Post Topic: USATT Rating Adjustment?
    Posted: 03/31/2022 at 2:51pm
Hi,

Why was this player not adjusted?

Player:Daniel Gong
Pre Tournament Rating:1985
Post Tournament Rating:2111


WinnerLoser
NameUSATT#NameUSATT#EventScore+/-Head to Head
Gong, Daniel (1985)90833
Yap, Connie
 (
1596
)
94145Open Singles5,4,30vs
Gong, Daniel (1985)90833
Seitz, Andrew
 (
1420
)
1168276Open Singles-9,4,6,30vs
Gong, Daniel (1985)90833
Milford, Jeff
 (
1822
)
220156Under 2200 Singles9,7,-8,72vs
Gong, Daniel (1985)90833
Elliott, Paul
 (
1937
)
32979Open Singles-Quarters7,9,-4,56vs
Gong, Daniel (1985)90833
Gibson, Ryan"TheNightmare"
 (
2178
)
89377Open Singles-Semis8,7,-6,-7,740vs
Gong, Daniel (1985)90833
Tran, Hiep
 (
2136
)
16107Open Singles-Finals8,9,-8,1030vs
Gong, Daniel (1985)90833
Li, Jiang
 (
1799
)
86652Under 2200 Singles-Quarters8,-7,4,82vs
Gong, Daniel (1985)90833
Elliott, Paul
 (
1937
)
32979Under 2200 Singles-Semis8,8,86vs
Gong, Daniel (1985)90833
Gibson, Ryan"TheNightmare"
 (
2178
)
89377Under 2200 Singles-Finals9,-9,3,-7,1040vs

Muhammad Ali reacted to light in 150 milliseconds, near the theoretical limit of human visual reaction time.
Back to Top
Sponsored Links


Back to Top
Lightspin View Drop Down
Super Member
Super Member


Joined: 07/11/2018
Location: USA
Status: Offline
Points: 368
Post Options Post Options   Thanks (0) Thanks(0)   Quote Lightspin Quote  Post ReplyReply Direct Link To This Post Posted: 03/31/2022 at 3:01pm
Sometimes the rating adjustment would be lower than the actual amount of points scored.  One tournament a 23xx guy dropped their rating down to 2000, beat 12 or so 2250 people and ended up with a rating close to 2600.  If they would have been readjusted they would be close to 2300 or so.  Other times the player doesn't get adjusted for some arbitrary reason.  I am not a fan of the process because of the "sometimes we will readjust and other times we won't" policy.

Maybe he didn't get readjusted because he didn't lose? One of the formulas has some sort of (Best win rating + worst loss rating)/2 component.  This can be exploited if you happen to play the open and lose only once to a 2700 player. 
Back to Top
Nesdin View Drop Down
Beginner
Beginner


Joined: 03/31/2022
Location: Seattle, USA
Status: Offline
Points: 2
Post Options Post Options   Thanks (0) Thanks(0)   Quote Nesdin Quote  Post ReplyReply Direct Link To This Post Posted: 03/31/2022 at 3:31pm
My understanding is that your new USATT rating equals (Rating of highest player you beat minus the rating of lowest rated player you lost) divided by 2 (but with some adjustments first in the 3 of earlier passes before the final 4th pass thru computer number crunching) 
This was years ago. May have changed now

Curious about how other rating systems like Ratings Central do this 


Edited by Nesdin - 03/31/2022 at 4:19pm
Back to Top
DonnOlsen View Drop Down
Gold Member
Gold Member
Avatar

Joined: 11/15/2008
Location: Maryland, USA
Status: Offline
Points: 1635
Post Options Post Options   Thanks (0) Thanks(0)   Quote DonnOlsen Quote  Post ReplyReply Direct Link To This Post Posted: 03/31/2022 at 5:11pm
Hi,

Thanks for the comments.  

The philosophy for the existence of the adjustment is to protect the higher rated players from losing an excessive number of points to a player that is demonstrably underrated.  Here we have the case that the higher rated players were not protected from this.

Given the above, the adjustment feature failed to fulfill its purpose.

The formula I'm aware of also includes a reflection of the worst loss in the calculation.  As this player did not lose, the formula lacks viability.  From my understanding, a rating adjustment feature under discussion here would need two formulas, one reflecting the worst loss and one reflecting this case of no loss in the tournament.

Thanks again. 
Muhammad Ali reacted to light in 150 milliseconds, near the theoretical limit of human visual reaction time.
Back to Top
Nesdin View Drop Down
Beginner
Beginner


Joined: 03/31/2022
Location: Seattle, USA
Status: Offline
Points: 2
Post Options Post Options   Thanks (0) Thanks(0)   Quote Nesdin Quote  Post ReplyReply Direct Link To This Post Posted: 03/31/2022 at 5:35pm
Originally posted by DonnOlsen DonnOlsen wrote:

Hi,

Thanks for the comments.  

The philosophy for the existence of the adjustment is to protect the higher rated players from losing an excessive number of points to a player that is demonstrably underrated.  Here we have the case that the higher rated players were not protected from this.

Given the above, the adjustment feature failed to fulfill its purpose.

The formula I'm aware of also includes a reflection of the worst loss in the calculation.  As this player did not lose, the formula lacks viability.  From my understanding, a rating adjustment feature under discussion here would need two formulas, one reflecting the worst loss and one reflecting this case of no loss in the tournament.

Thanks again. 

Same as what I said ?

(Rating of highest player you beat minus the rating of lowest rated player you lost) /2

Rating of lowest rated player you lost = Worst loss 

No loss > Rating of highest player you beat

I don't see a need for two formulas  ??
 

Back to Top
pongfugrasshopper View Drop Down
Platinum Member
Platinum Member
Avatar

Joined: 03/22/2015
Location: USA
Status: Offline
Points: 2453
Post Options Post Options   Thanks (0) Thanks(0)   Quote pongfugrasshopper Quote  Post ReplyReply Direct Link To This Post Posted: 03/31/2022 at 7:25pm
If you click on the player's Final Rating for that tournament, it gives the message:

Ratings explainer is temporarily unavailable. Please check back later.

So it looks like maybe a glitch in the Matrix?  I'd suggest waiting a few days for them to sort things out.

Anyway, here's the link for the Ratings System Explanation (it does account for all wins/all losses):

Back to Top
Gordy View Drop Down
Silver Member
Silver Member


Joined: 12/04/2011
Location: United States
Status: Offline
Points: 672
Post Options Post Options   Thanks (0) Thanks(0)   Quote Gordy Quote  Post ReplyReply Direct Link To This Post Posted: 03/31/2022 at 7:45pm
It's actually something called the "median implied rating" when dealing with a player with all wins and it dates back quite a while. 

What happened here is the player had several good wins that triggered the adjustment algorithm, but since he didn't have any losses, the system looked at the median of the ratings wins. In this case, the median of his wins was 1937 - lower than his pre-tournament rating, so no adjustment was made.

"If a player has either, all wins, or all losses, the Adjusted Rating is derived by taking the median implied rating for all of the player’s games. The implied rating is calculated using each of the opponents’ Pre-Tournament Ratings, and the Rating Chart above."

In 2016, the algorithm was changed to recognize a case like this (where the PASS1 points gain was greater than 75, all wins, and the median rating was less than the pre-tournament rating), an adjustment would be made to the pre-tournament rating + PASS1 points...but somewhere or some time it was changed back. I am not sure when.
JOOLA Nobilis / SP (FH)/ Rhyzm Tech (BH)
Back to Top
stiltt View Drop Down
Assistant Admin
Assistant Admin
Avatar

Joined: 07/15/2007
Location: Seattle, WA
Status: Offline
Points: 165
Post Options Post Options   Thanks (0) Thanks(0)   Quote stiltt Quote  Post ReplyReply Direct Link To This Post Posted: 03/31/2022 at 10:59pm
I am curious about what his rating would have beed post tourney had he lost one of his 2 matches v. Ryan Gibsin 2178.
Back to Top
DonnOlsen View Drop Down
Gold Member
Gold Member
Avatar

Joined: 11/15/2008
Location: Maryland, USA
Status: Offline
Points: 1635
Post Options Post Options   Thanks (0) Thanks(0)   Quote DonnOlsen Quote  Post ReplyReply Direct Link To This Post Posted: 03/31/2022 at 11:19pm
I don't see a need for two formulas  ??

The reason for the need for two formulas is that the current state of processing with a no-loss situation of the under-rated player fails to protect the higher rated player from egregious point losses, as is precisely the purpose of the adjustment feature in the rating system to provide such protection.  

The system, in this case, failed to protect the higher rated players, purely because the under-rated player did not have a loss, a condition that the higher rated players are not responsible for, yet, the higher rated players are penalized.

The previously higher rated players now have considerably lower ratings than would be the case if the adjustment had been made.  This, prima facie, results in them carrying into the next tournament a rating that does not reflect their actual playing level, which is the primary purpose of the rating system to accurately represent.  This erroneous state can then potentially distort the integrity of the next entered previously higher rated players' tournament.

Thanks.
Muhammad Ali reacted to light in 150 milliseconds, near the theoretical limit of human visual reaction time.
Back to Top
Gordy View Drop Down
Silver Member
Silver Member


Joined: 12/04/2011
Location: United States
Status: Offline
Points: 672
Post Options Post Options   Thanks (0) Thanks(0)   Quote Gordy Quote  Post ReplyReply Direct Link To This Post Posted: 04/01/2022 at 3:20am
You can figure out what would have happened with one loss...

Let's say he lost one of the matches to the 2178...His adjustment would have been (2178 + 2178 )/ 2 =2178, then (2178 + 1985) /2 = 2082

Had he lost to the 1822, his adjustment would have been (2178 + 1822)/2 = 2000, then (2000+1985)/2 = 1992


JOOLA Nobilis / SP (FH)/ Rhyzm Tech (BH)
Back to Top
pongfugrasshopper View Drop Down
Platinum Member
Platinum Member
Avatar

Joined: 03/22/2015
Location: USA
Status: Offline
Points: 2453
Post Options Post Options   Thanks (0) Thanks(0)   Quote pongfugrasshopper Quote  Post ReplyReply Direct Link To This Post Posted: 04/01/2022 at 6:46am
Originally posted by Gordy Gordy wrote:

You can figure out what would have happened with one loss...

Let's say he lost one of the matches to the 2178...His adjustment would have been (2178 + 2178 )/ 2 =2178, then (2178 + 1985) /2 = 2082

Had he lost to the 1822, his adjustment would have been (2178 + 1822)/2 = 2000, then (2000+1985)/2 = 1992


Doesn't it seem odd that he would have been adjusted higher had he *lost* to say 2178?  It may or may not matter to him, but I suspect it matters to some of his opponents.  The fact that he gained 126 points in the 1st pass should trigger that he needs to be adjusted higher, and the question then becomes how much higher. But in this case the middle value says nope, no adjustment.
Back to Top
DonnOlsen View Drop Down
Gold Member
Gold Member
Avatar

Joined: 11/15/2008
Location: Maryland, USA
Status: Offline
Points: 1635
Post Options Post Options   Thanks (0) Thanks(0)   Quote DonnOlsen Quote  Post ReplyReply Direct Link To This Post Posted: 04/01/2022 at 7:33am
"If a player has either, all wins, or all losses, the Adjusted Rating is derived by taking the median implied rating for all of the player’s games. The implied rating is calculated using each of the opponents’ Pre-Tournament Ratings, and the Rating Chart above."

An "all losses" condition would not invoke an Adjusted Rating.

__________________________________________________________

the Adjusted Rating is derived by taking the median implied rating for all of the player’s games.

This is also not correct, as it moves the metric target to games, not matches.  Matches are used, not games within the matches.

Gee, no wonder the confusion.

Thanks.



Edited by DonnOlsen - 04/01/2022 at 7:37am
Muhammad Ali reacted to light in 150 milliseconds, near the theoretical limit of human visual reaction time.
Back to Top
pongfugrasshopper View Drop Down
Platinum Member
Platinum Member
Avatar

Joined: 03/22/2015
Location: USA
Status: Offline
Points: 2453
Post Options Post Options   Thanks (0) Thanks(0)   Quote pongfugrasshopper Quote  Post ReplyReply Direct Link To This Post Posted: 04/01/2022 at 8:05am
Originally posted by Gordy Gordy wrote:

It's actually something called the "median implied rating" when dealing with a player with all wins and it dates back quite a while. 

What happened here is the player had several good wins that triggered the adjustment algorithm, but since he didn't have any losses, the system looked at the median of the ratings wins. In this case, the median of his wins was 1937 - lower than his pre-tournament rating, so no adjustment was made.

"If a player has either, all wins, or all losses, the Adjusted Rating is derived by taking the median implied rating for all of the player’s games. The implied rating is calculated using each of the opponents’ Pre-Tournament Ratings, and the Rating Chart above."

In 2016, the algorithm was changed to recognize a case like this (where the PASS1 points gain was greater than 75, all wins, and the median rating was less than the pre-tournament rating), an adjustment would be made to the pre-tournament rating + PASS1 points...but somewhere or some time it was changed back. I am not sure when.
The actual text on Ratings System Explanation page is as you described.  So it appears as if there's a discrepancy between what's published and what's actually used in the algorithm.  I remember complaining about this way back when and you fixed this (kudos to you!).  Sadly, it looks like we've reverted back to the land of confusion.
Back to Top
Paul E View Drop Down
Beginner
Beginner


Joined: 04/29/2022
Location: Virginia
Status: Offline
Points: 1
Post Options Post Options   Thanks (0) Thanks(0)   Quote Paul E Quote  Post ReplyReply Direct Link To This Post Posted: 04/29/2022 at 10:02pm
Hey Donn,

I've sent message after message to USATT about this. It took a while to break through, but I finally got (I think at least) some real help after emailing the CEO Virginia Sung. She put me in touch with one of their software maintainers a couple days ago, and he said he would look into the rating discrepancy for Daniel Gong. Apparently the ratings explainer in Simply Complete has been down for a long while. USATT is supposed to have it fixed soon. But that's what they keep on saying. And after emailing five different people, I still have yet to receive an answer from the folks (USATT) that created the rating system. 

According to my calculation, he should be adjusted to 2111 with a final rating of 2157. My only worry is that after 60 days, according to USATT, ratings corrections cannot be made. It's already been over a month.


Edited by Paul E - 04/29/2022 at 10:04pm
Back to Top
David Marcus View Drop Down
Super Member
Super Member
Avatar

Joined: 10/28/2010
Location: Somerville, MA
Status: Offline
Points: 132
Post Options Post Options   Thanks (0) Thanks(0)   Quote David Marcus Quote  Post ReplyReply Direct Link To This Post Posted: 05/15/2022 at 3:00pm
Originally posted by Nesdin Nesdin wrote:

Curious about how other rating systems like Ratings Central do this

The Ratings Central algorithm is quite different from the USATT algorithm. One way to look at the Ratings Central algorithm is that every opponent is "adjusted" when calculating your rating. I.e., when calculating your rating, the Ratings Central algorithm takes into account how your opponent did in the same tournament, rather than only using their pre-tournament rating. For details, see https://www.ratingscentral.com/HowItWorks.php .
Back to Top
alford View Drop Down
Beginner
Beginner


Joined: 05/15/2022
Location: Iowa City
Status: Offline
Points: 15
Post Options Post Options   Thanks (0) Thanks(0)   Quote alford Quote  Post ReplyReply Direct Link To This Post Posted: 05/15/2022 at 9:19pm
Originally posted by David Marcus David Marcus wrote:

Originally posted by Nesdin Nesdin wrote:

Curious about how other rating systems like Ratings Central do this

The Ratings Central algorithm is quite different from the USATT algorithm. One way to look at the Ratings Central algorithm is that every opponent is "adjusted" when calculating your rating. I.e., when calculating your rating, the Ratings Central algorithm takes into account how your opponent did in the same tournament, rather than only using their pre-tournament rating. For details, see https://www.ratingscentral.com/HowItWorks.php .

I beg to differ. I am not exactly a big fan of USATT & quit being a USATT member long ago, and  USATT League rating system may be a simpler one pass system, but as far as I know,  USATT tournament rating system is a 4 pass algorithm where it does indeed take into the account the ratings of other players in a given tournament. 

Would like to hear about Canadian, German & any other national rating methods.


Edited by alford - 05/15/2022 at 9:23pm
Back to Top
David Marcus View Drop Down
Super Member
Super Member
Avatar

Joined: 10/28/2010
Location: Somerville, MA
Status: Offline
Points: 132
Post Options Post Options   Thanks (0) Thanks(0)   Quote David Marcus Quote  Post ReplyReply Direct Link To This Post Posted: 05/15/2022 at 11:30pm
Originally posted by alford alford wrote:

I beg to differ. I am not exactly a big fan of USATT & quit being a USATT member long ago, and  USATT League rating system may be a simpler one pass system, but as far as I know,  USATT tournament rating system is a 4 pass algorithm where it does indeed take into the account the ratings of other players in a given tournament. 

Would like to hear about Canadian, German & any other national rating methods.

I didn't say USATT doesn't try to take into account how your opponents do. That is the whole point of the adjustments. But, it doesn't do this in a uniform way. The Ratings Central way of doing adjustments is very different from the USATT way. The USATT adjustments make more sense than the USATT rating chart. The main problem with the USATT system is the lack of memory.

Last I checked, the Canadian system was sort of like the USATT, but more inflationary. Don't know if they've changed it recently.

The Austrian Table Tennis Association, Table Tennis Australia, and ITTF-Oceania use Ratings Central for their official ratings/rankings.
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 0.063 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.