Alex Table Tennis - MyTableTennis.NET Homepage
  New Posts New Posts RSS Feed - Usatt Rating (Does this rating look correct)
  FAQ FAQ  Forum Search   Events   Register Register  Login Login

Usatt Rating (Does this rating look correct)

 Post Reply Post Reply Page  <1234 5>
Author
pongfugrasshopper View Drop Down
Gold Member
Gold Member
Avatar

Joined: 03/22/2015
Location: USA
Status: Offline
Points: 1029
Post Options Post Options   Thanks (0) Thanks(0)   Quote pongfugrasshopper Quote  Post ReplyReply Direct Link To This Post Posted: 02/06/2019 at 2:46pm
Originally posted by tom tom wrote:

Originally posted by pongfugrasshopper pongfugrasshopper wrote:

Originally posted by Vince64 Vince64 wrote:

Originally posted by tom tom wrote:

but if my applcation is right then the rule does not make sense, because instead of his worst loss was to a 2550 player it was to a 2800 player his pass 2 adjustment would have yield him a much higher rating.  a loss is a loss - why would losing to a higher rated player give you higher points than if you lost to a less higher rated player?
Where are you coming up with a loss to a 2800 player? The OP played 3 matches. He beat 2186 and 2339 and lost to 2550. 
I think he's speaking hypothetically.  Why should your rating increase when losing to a 2550 vs. losing to a 2800?  That's a good point.  
Pongfu, you are at an even 1000 points.  If I were you I would never post again just to keep it at that special #
Thanks, I didn't even realize that until you mentioned it.  And to think, I didn't even make the active members list! LOL
Back to Top
Pushblocker View Drop Down
Gold Member
Gold Member
Avatar

Joined: 12/09/2009
Location: Florida
Status: Offline
Points: 1819
Post Options Post Options   Thanks (0) Thanks(0)   Quote Pushblocker Quote  Post ReplyReply Direct Link To This Post Posted: 02/06/2019 at 7:31pm
Originally posted by pongfugrasshopper pongfugrasshopper wrote:

Originally posted by Pushblocker Pushblocker wrote:

I looked up the match and made the calculation.. Based on my understanding of the rating system, you should have ended up at 2320...

(Pass1 result (2147) + (best win + worst loss / 2)) / 2
(2147 +((2598 + 2339) / 2)) / 2 = 2307.25 (2307 adjusted rating)
Then you would have gained 10 points against the 2337 and 3 points against the 2186... 2320 total

Maybe they changed the "big adjustment" threshold to 76???
 
Since the point gain is at least 75 from the initial calculation, you need to use the PASS2 calculation which is what heavyspin did above with slight correction by Vince64.
I disagree.. For gains between 50 and 74 points, pass 1 rating is supposed to be used and for 75+, the process I outlined is supposed to be used..


Edited by Pushblocker - 02/06/2019 at 7:32pm
2010 Florida State Champion

Dr. Neubauer Firewall Plus Blade with DHS G666 1.5mm on forehand Giant Dragon Talon OX with removed glue sheet on backhand
Back to Top
pongfugrasshopper View Drop Down
Gold Member
Gold Member
Avatar

Joined: 03/22/2015
Location: USA
Status: Offline
Points: 1029
Post Options Post Options   Thanks (0) Thanks(0)   Quote pongfugrasshopper Quote  Post ReplyReply Direct Link To This Post Posted: 02/06/2019 at 7:53pm
Originally posted by Pushblocker Pushblocker wrote:

Originally posted by pongfugrasshopper pongfugrasshopper wrote:

Originally posted by Pushblocker Pushblocker wrote:

I looked up the match and made the calculation.. Based on my understanding of the rating system, you should have ended up at 2320...

(Pass1 result (2147) + (best win + worst loss / 2)) / 2
(2147 +((2598 + 2339) / 2)) / 2 = 2307.25 (2307 adjusted rating)
Then you would have gained 10 points against the 2337 and 3 points against the 2186... 2320 total

Maybe they changed the "big adjustment" threshold to 76???
 
Since the point gain is at least 75 from the initial calculation, you need to use the PASS2 calculation which is what heavyspin did above with slight correction by Vince64.
I disagree.. For gains between 50 and 74 points, pass 1 rating is supposed to be used and for 75+, the process I outlined is supposed to be used..

I think you may be right about the big adjustment threshold.  Another possibility that I'm speculating about is that perhaps they've changed the algorithm a bit so that we don't run into the problem that tom mentioned.  Perhaps if worst loss is significantly above best win, then it reverts back to no adjustment.  

I remember a few years back, a number of forum members, including myself, were arguing that the whole ratings process should be transparent like it had been a long time ago.  Gordon was in full agreement, and he made it a point to put the algorithm back onto the website.  However, it seems that whether intentionally or not, that transparency seems to have eroded.  If it's now the case that the full algorithm is now hidden and is only known to usatt as taczkid discovered, then that's a disappointing step backwards.
Back to Top
Pushblocker View Drop Down
Gold Member
Gold Member
Avatar

Joined: 12/09/2009
Location: Florida
Status: Offline
Points: 1819
Post Options Post Options   Thanks (0) Thanks(0)   Quote Pushblocker Quote  Post ReplyReply Direct Link To This Post Posted: 02/07/2019 at 6:47am
Originally posted by pongfugrasshopper pongfugrasshopper wrote:

Originally posted by Pushblocker Pushblocker wrote:

Originally posted by pongfugrasshopper pongfugrasshopper wrote:

Originally posted by Pushblocker Pushblocker wrote:

I looked up the match and made the calculation.. Based on my understanding of the rating system, you should have ended up at 2320...

(Pass1 result (2147) + (best win + worst loss / 2)) / 2
(2147 +((2598 + 2339) / 2)) / 2 = 2307.25 (2307 adjusted rating)
Then you would have gained 10 points against the 2337 and 3 points against the 2186... 2320 total

Maybe they changed the "big adjustment" threshold to 76???
 
Since the point gain is at least 75 from the initial calculation, you need to use the PASS2 calculation which is what heavyspin did above with slight correction by Vince64.
I disagree.. For gains between 50 and 74 points, pass 1 rating is supposed to be used and for 75+, the process I outlined is supposed to be used..

I think you may be right about the big adjustment threshold.  Another possibility that I'm speculating about is that perhaps they've changed the algorithm a bit so that we don't run into the problem that tom mentioned.  Perhaps if worst loss is significantly above best win, then it reverts back to no adjustment.  

I remember a few years back, a number of forum members, including myself, were arguing that the whole ratings process should be transparent like it had been a long time ago.  Gordon was in full agreement, and he made it a point to put the algorithm back onto the website.  However, it seems that whether intentionally or not, that transparency seems to have eroded.  If it's now the case that the full algorithm is now hidden and is only known to usatt as taczkid discovered, then that's a disappointing step backwards.
They must have changed the algorithm. I'm certain of the calculation as I have used it many times. It used to be posted on the USATT webpage but at one point, they removed the adjustment calculation.
The way "BIG" adjustments are calculated has not changed but it appears that the threshold has changed. It always used to be 75 points to get a big adjustment (considering best win and worst loss) and 50 points to get a normal adjustment. I know that they changed the minimum points required to get an adjustment from 51 to 50 points but I was not aware of any changes to the process for larger gains.
2010 Florida State Champion

Dr. Neubauer Firewall Plus Blade with DHS G666 1.5mm on forehand Giant Dragon Talon OX with removed glue sheet on backhand
Back to Top
mentortt View Drop Down
Member
Member


Joined: 07/17/2018
Location: NY
Status: Offline
Points: 40
Post Options Post Options   Thanks (0) Thanks(0)   Quote mentortt Quote  Post ReplyReply Direct Link To This Post Posted: 02/07/2019 at 8:38am
Thanks for Pushblocker for information. I have written a small script to calculate rating automatically. 

I think the issue is >75 or >=75. For OP 's case, if >75, the adjusted rating will be 2147 and the new rating will be 2200. If >=75 (USATT rating explanation), the adjusted rating will be 2295(rounded), the new rating will be 2312.

 

One thing to note: In USATT website the following sentence is incorrect: "the Adjusted Rating is derived by taking the average of the player’s Pre-Tournament Rating, and the average of the player’s best win and worst loss." What Pushblocker's example is correct: It is not Pre-tournament rating. It should be first pass rating. I have confirmed this case by other person's rating changes in latest tournament. I have also found a 2001 paper which also confirms that. The USATT rating explanation needs to be corrected.






Edited by mentortt - 02/07/2019 at 9:45am
Back to Top
tom View Drop Down
Gold Member
Gold Member


Joined: 11/18/2013
Location: canada
Status: Offline
Points: 1477
Post Options Post Options   Thanks (0) Thanks(0)   Quote tom Quote  Post ReplyReply Direct Link To This Post Posted: 02/07/2019 at 10:08am
Originally posted by mentortt mentortt wrote:

Thanks for Pushblocker for information. I have written a small script to calculate rating automatically. 

I think the issue is >75 or >=75. For OP 's case, if >75, the adjusted rating will be 2147 and the new rating will be 2200. If >=75 (USATT rating explanation), the adjusted rating will be 2295(rounded), the new rating will be 2312.

 

One thing to note: In USATT website the following sentence is incorrect: "the Adjusted Rating is derived by taking the average of the player’s Pre-Tournament Rating, and the average of the player’s best win and worst loss." What Pushblocker's example is correct: It is not Pre-tournament rating. It should be first pass rating. I have confirmed this case by other person's rating changes in latest tournament. I have also found a 2001 paper which also confirms that. The USATT rating explanation needs to be corrected.




this subject needs to be addressed by the USATT urgently, if the stuff that has been posted about their rating system is even half true.  Imagine the rating system being opaque to its members who care about it. It's like having a government election that the voters can't find out how it works.
Back to Top
taczkid View Drop Down
Super Member
Super Member
Avatar

Joined: 05/19/2016
Location: ILLINOIS
Status: Offline
Points: 345
Post Options Post Options   Thanks (0) Thanks(0)   Quote taczkid Quote  Post ReplyReply Direct Link To This Post Posted: 02/07/2019 at 11:04am
All these explanations make sense to me and to a lot of users on this forum, where as we mostly agree that my rating based on USATT rules written on their website should be around 2300 + -, but USATT SECRET ALGORITHMS say something else LOLLOLLOLLOLLOLLOLLOL
Back to Top
Pushblocker View Drop Down
Gold Member
Gold Member
Avatar

Joined: 12/09/2009
Location: Florida
Status: Offline
Points: 1819
Post Options Post Options   Thanks (0) Thanks(0)   Quote Pushblocker Quote  Post ReplyReply Direct Link To This Post Posted: 02/07/2019 at 11:12am
Originally posted by taczkid taczkid wrote:

All these explanations make sense to me and to a lot of users on this forum, where as we mostly agree that my rating based on USATT rules written on their website should be around 2300 + -, but USATT SECRET ALGORITHMS say something else LOLLOLLOLLOLLOLLOLLOL

I'm certain that the calculation I have for the "big" adjustment is correct as I correctly calculated a friends adjustment last June.. However, he gained more than 100 points in pass 1, so it is hard to tell if they just shifted the minimum gain in pass 1 required to get into the calculation that considers best win and worst loss.. My guess is that they increased the threshold for that type of adjustment to be above 75 points gained in pass 1.


Edited by Pushblocker - 02/07/2019 at 11:14am
2010 Florida State Champion

Dr. Neubauer Firewall Plus Blade with DHS G666 1.5mm on forehand Giant Dragon Talon OX with removed glue sheet on backhand
Back to Top
Pushblocker View Drop Down
Gold Member
Gold Member
Avatar

Joined: 12/09/2009
Location: Florida
Status: Offline
Points: 1819
Post Options Post Options   Thanks (0) Thanks(0)   Quote Pushblocker Quote  Post ReplyReply Direct Link To This Post Posted: 02/07/2019 at 11:28am
Here is the details on the adjustment that a friend of mine got last June:

Initial Rating: 2191

W
1790
2137
2389
2511
2262

L
2530
2466

First pass:
wins: 1790 +0, 2137 +6, 2389 +40, 2511 +50, 2262 +16
losses:  2530 -0, 2466 -0
6+40+50+16 = 112 points gained

Pass 1 rating: 2191 + 112 = 2303
Adjustment calculation:
(2303 + (2511+2466/2) ) / 2  = 2395.75 (rating system rounds down, so he was adjusted to 2395)

Then the wins/losses were re-applied with the adjusted ratings (his and his opponents)

His final rating ended up being 2426 (would have been 2423 but one of his opponents also got an adjustment and he picked up 3 extra points to get to 2426)

Final Pass:

2395
W    
1790 +0 
2137 +0
2389 +8
2511 +25
2347 +6
L
2530 -3
2466 -5

2395 + 8 + 25 + 6 - 3 - 5 = 2426



Edited by Pushblocker - 02/07/2019 at 11:38am
2010 Florida State Champion

Dr. Neubauer Firewall Plus Blade with DHS G666 1.5mm on forehand Giant Dragon Talon OX with removed glue sheet on backhand
Back to Top
qpskfec View Drop Down
Super Member
Super Member


Joined: 07/28/2011
Status: Offline
Points: 357
Post Options Post Options   Thanks (0) Thanks(0)   Quote qpskfec Quote  Post ReplyReply Direct Link To This Post Posted: 02/07/2019 at 11:39am
According to larrytt's latest blog, there will be an election soon for a board of director's position at usatt. Members will vote on who gets this slot. http://tabletenniscoaching.com/node/3037

If I was a usatt member I would ask all the candidates to state their position on secrecy and not vote for any candidate that wants to keep the algorithm secret.

Back to Top
tom View Drop Down
Gold Member
Gold Member


Joined: 11/18/2013
Location: canada
Status: Offline
Points: 1477
Post Options Post Options   Thanks (0) Thanks(0)   Quote tom Quote  Post ReplyReply Direct Link To This Post Posted: 02/07/2019 at 12:19pm
Originally posted by qpskfec qpskfec wrote:

According to larrytt's latest blog, there will be an election soon for a board of director's position at usatt. Members will vote on who gets this slot. http://tabletenniscoaching.com/node/3037

If I was a usatt member I would ask all the candidates to state their position on secrecy and not vote for any candidate that wants to keep the algorithm secret.

What a joke ! - keeping the algorithm secret 
Back to Top
mentortt View Drop Down
Member
Member


Joined: 07/17/2018
Location: NY
Status: Offline
Points: 40
Post Options Post Options   Thanks (0) Thanks(0)   Quote mentortt Quote  Post ReplyReply Direct Link To This Post Posted: 02/07/2019 at 12:24pm
@Pushblocker. For your case with 2426, my script can calculate result accurately. I test multiple cases and one case cannot find the answer if you can shed some lights. 

Begin=1673
won= 1451,1411,1347,1208,1897,1579,1525,1320,1260,1317,1675,1724,1710,1607,1734
lost = 1846,1735,1934,1739

USATT results: adjust rating: 1758.
But my results: 1788. Based on rating change 88, it will go to pass2.  (1761+(1897+1735)/2)/2 = 1788.5. So the adjusted rating will be 1788. 

Anything wrong here? Pure algorithm check. I want to make everything correctly.



Back to Top
pgpg View Drop Down
Gold Member
Gold Member
Avatar

Joined: 11/18/2013
Location: United States
Status: Offline
Points: 1089
Post Options Post Options   Thanks (0) Thanks(0)   Quote pgpg Quote  Post ReplyReply Direct Link To This Post Posted: 02/07/2019 at 12:27pm
Originally posted by mentortt mentortt wrote:

@Pushblocker. For your case with 2426, my script can calculate result accurately. I test multiple cases and one case cannot find the answer if you can shed some lights. 

Begin=1673
won= 1451,1411,1347,1208,1897,1579,1525,1320,1260,1317,1675,1724,1710,1607,1734
lost = 1846,1735,1934,1739

USATT results: adjust rating: 1758.
But my results: 1788. Based on rating change 88, it will go to pass2.  (1761+(1897+1735)/2)/2 = 1788.5. So the adjusted rating will be 1788. 

Anything wrong here? Pure algorithm check. I want to make everything correctly.


With that many opponents, I bet that some of them were adjusted too, so you need to run your algorithm on ALL participants and keep track of their adjustments as well. Which makes debugging your code a small scale nightmare.
USATT: ~1900
Donic Defplay - AK47 blue - Dtecs OX
Back to Top
mentortt View Drop Down
Member
Member


Joined: 07/17/2018
Location: NY
Status: Offline
Points: 40
Post Options Post Options   Thanks (0) Thanks(0)   Quote mentortt Quote  Post ReplyReply Direct Link To This Post Posted: 02/07/2019 at 12:32pm
Originally posted by pgpg pgpg wrote:

Originally posted by mentortt mentortt wrote:

@Pushblocker. For your case with 2426, my script can calculate result accurately. I test multiple cases and one case cannot find the answer if you can shed some lights. 

Begin=1673
won= 1451,1411,1347,1208,1897,1579,1525,1320,1260,1317,1675,1724,1710,1607,1734
lost = 1846,1735,1934,1739

USATT results: adjust rating: 1758.
But my results: 1788. Based on rating change 88, it will go to pass2.  (1761+(1897+1735)/2)/2 = 1788.5. So the adjusted rating will be 1788. 

Anything wrong here? Pure algorithm check. I want to make everything correctly.


With that many opponents, I bet that some of them were adjusted too, so you need to run your algorithm on ALL participants and keep track of their adjustments as well. Which makes debugging your code a small scale nightmare.

No. All other players are already adjusted. This is the only case my script get wrong answer. Hope some person can find problems.
Back to Top
taczkid View Drop Down
Super Member
Super Member
Avatar

Joined: 05/19/2016
Location: ILLINOIS
Status: Offline
Points: 345
Post Options Post Options   Thanks (0) Thanks(0)   Quote taczkid Quote  Post ReplyReply Direct Link To This Post Posted: 02/07/2019 at 1:02pm
I just looked up player Member id 88997 and looked at his rating, he got adjusted nicely,,, he had similar results, 2 good wins so USATT adjusted his rating to average between the two wins he had... in my scenario they did not do that... or I would  have to have 2262 afjusted to instead of 2147 they adjusted me to... All this is very interesting to me why for example this player 88997 who like myself earned 75+ pts did get adjusted totally diferent way then my case...
I did post this question and asked for answer from USATT today, will share what they reply...
 
Back to Top
Pushblocker View Drop Down
Gold Member
Gold Member
Avatar

Joined: 12/09/2009
Location: Florida
Status: Offline
Points: 1819
Post Options Post Options   Thanks (0) Thanks(0)   Quote Pushblocker Quote  Post ReplyReply Direct Link To This Post Posted: 02/07/2019 at 1:43pm
Originally posted by mentortt mentortt wrote:

@Pushblocker. For your case with 2426, my script can calculate result accurately. I test multiple cases and one case cannot find the answer if you can shed some lights. 

Begin=1673
won= 1451,1411,1347,1208,1897,1579,1525,1320,1260,1317,1675,1724,1710,1607,1734
lost = 1846,1735,1934,1739

USATT results: adjust rating: 1758.
But my results: 1788. Based on rating change 88, it will go to pass2.  (1761+(1897+1735)/2)/2 = 1788.5. So the adjusted rating will be 1788. 

Anything wrong here? Pure algorithm check. I want to make everything correctly.



Please either post or PM me the tournament in question and I will do my research.. I need to see all the other player's initial rating (prior to any potential adjustments etc.) as also all unrated players you played..

2010 Florida State Champion

Dr. Neubauer Firewall Plus Blade with DHS G666 1.5mm on forehand Giant Dragon Talon OX with removed glue sheet on backhand
Back to Top
Pushblocker View Drop Down
Gold Member
Gold Member
Avatar

Joined: 12/09/2009
Location: Florida
Status: Offline
Points: 1819
Post Options Post Options   Thanks (0) Thanks(0)   Quote Pushblocker Quote  Post ReplyReply Direct Link To This Post Posted: 02/07/2019 at 1:46pm
Originally posted by Pushblocker Pushblocker wrote:

Originally posted by mentortt mentortt wrote:

@Pushblocker. For your case with 2426, my script can calculate result accurately. I test multiple cases and one case cannot find the answer if you can shed some lights. 

Begin=1673
won= 1451,1411,1347,1208,1897,1579,1525,1320,1260,1317,1675,1724,1710,1607,1734
lost = 1846,1735,1934,1739

USATT results: adjust rating: 1758.
But my results: 1788. Based on rating change 88, it will go to pass2.  (1761+(1897+1735)/2)/2 = 1788.5. So the adjusted rating will be 1788. 

Anything wrong here? Pure algorithm check. I want to make everything correctly.



Please either post or PM me the tournament in question and I will do my research.. I need to see all the other player's initial rating (prior to any potential adjustments etc.) as also all unrated players you played..

never mind.. found it..
2010 Florida State Champion

Dr. Neubauer Firewall Plus Blade with DHS G666 1.5mm on forehand Giant Dragon Talon OX with removed glue sheet on backhand
Back to Top
Pushblocker View Drop Down
Gold Member
Gold Member
Avatar

Joined: 12/09/2009
Location: Florida
Status: Offline
Points: 1819
Post Options Post Options   Thanks (0) Thanks(0)   Quote Pushblocker Quote  Post ReplyReply Direct Link To This Post Posted: 02/07/2019 at 2:03pm
Originally posted by mentortt mentortt wrote:

@Pushblocker. For your case with 2426, my script can calculate result accurately. I test multiple cases and one case cannot find the answer if you can shed some lights. 

Begin=1673
won= 1451,1411,1347,1208,1897,1579,1525,1320,1260,1317,1675,1724,1710,1607,1734
lost = 1846,1735,1934,1739

USATT results: adjust rating: 1758.
But my results: 1788. Based on rating change 88, it will go to pass2.  (1761+(1897+1735)/2)/2 = 1788.5. So the adjusted rating will be 1788. 

Anything wrong here? Pure algorithm check. I want to make everything correctly.




The 1739 player that he lost to had a pre-adjustment rating of 1629 which is used to determine adjustments.. I highlighted the ratings used for each pass

So, the processing was like this:

number behind / is post adjustment rating of opponent

1673
W
1451 +1
1411 +0
1347+0
1208 +0
1897 +45
1579 +4
1525 +2
1320 +0
1088/1260 +0
1317 +0
1675 +8
1724 +13
1710 +10
1607 +5
1734 +13
L
1846 -2
1735 -6
1934 -0
1629/1739 -13


Wins: 1+45+4+2+8+13+10+5+13 = 101
losses: 6+2+13 = 21
Total points gained 101 - 21 = 80

pass 1 rating 1673 + 80 = 1753
pass 2 Adjustment calculation: 1753 + ((1897+1629/2)/2) = 1758 adjusted rating
Pass 3 final rating


1347+0
1208 +0
1897 +30
1579 +2
1525 +1
1320 +0
1088/1260 +0
1317 +0
1675 +4
1724 +7
1710 +6
1607 +2
1734 +7L
1846 -2
1735 -10
1934 -2
1629/1739 -10

Final Rating = 1792

Unlike the rating in the OP, the rating in this case was calculated 100% accurate


Edited by Pushblocker - 02/07/2019 at 2:05pm
2010 Florida State Champion

Dr. Neubauer Firewall Plus Blade with DHS G666 1.5mm on forehand Giant Dragon Talon OX with removed glue sheet on backhand
Back to Top
mentortt View Drop Down
Member
Member


Joined: 07/17/2018
Location: NY
Status: Offline
Points: 40
Post Options Post Options   Thanks (0) Thanks(0)   Quote mentortt Quote  Post ReplyReply Direct Link To This Post Posted: 02/07/2019 at 2:14pm
Quote

The 1739 player that he lost to had a pre-adjustment rating of 1629 which is used to determine adjustments.. I highlighted the ratings used for each pass


Excellent! So the first pass is using opponent pre-adjust rating. The second pass using opponent adjusted rating. correct?  New information to me. Thanks a lot!


Edited by mentortt - 02/07/2019 at 2:16pm
Back to Top
Pushblocker View Drop Down
Gold Member
Gold Member
Avatar

Joined: 12/09/2009
Location: Florida
Status: Offline
Points: 1819
Post Options Post Options   Thanks (1) Thanks(1)   Quote Pushblocker Quote  Post ReplyReply Direct Link To This Post Posted: 02/07/2019 at 2:18pm
Originally posted by mentortt mentortt wrote:

Quote

The 1739 player that he lost to had a pre-adjustment rating of 1629 which is used to determine adjustments.. I highlighted the ratings used for each pass


Excellent! So the first pass is using opponent pre-adjust rating. The second pass using opponent adjusted rating. correct?  New information to me. Thanks a lot!

Well, to be exact, it's the 3rd and final pass.. Pass 2 determines adjustments and Pass 3 determines the final rating.. Yes, the adjusted rating of your opponents is being used for the 3rd pass and the pre-adjustment rating for the first pass as also for pass 2 which determines adjustments. 

2010 Florida State Champion

Dr. Neubauer Firewall Plus Blade with DHS G666 1.5mm on forehand Giant Dragon Talon OX with removed glue sheet on backhand
Back to Top
Pushblocker View Drop Down
Gold Member
Gold Member
Avatar

Joined: 12/09/2009
Location: Florida
Status: Offline
Points: 1819
Post Options Post Options   Thanks (0) Thanks(0)   Quote Pushblocker Quote  Post ReplyReply Direct Link To This Post Posted: 02/07/2019 at 2:22pm
One issue with the rating system is also with unrated players.. Let's say a unrated player beats a 1400 and loses to a 1600 player.. He will get a 1500 initial rating.. However, what if the 1400 player got an adjustment to 1700, theoretically, the new player's initial rating should have been 1650 and not 1500. However, for initial rating, pre-adjustment ratings of opponents are being used which can lead to incorrect ratings if either the best win or worst loss of the player is against someone who was very underrated..
2010 Florida State Champion

Dr. Neubauer Firewall Plus Blade with DHS G666 1.5mm on forehand Giant Dragon Talon OX with removed glue sheet on backhand
Back to Top
Gordy View Drop Down
Silver Member
Silver Member


Joined: 12/04/2011
Location: United States
Status: Offline
Points: 648
Post Options Post Options   Thanks (0) Thanks(0)   Quote Gordy Quote  Post ReplyReply Direct Link To This Post Posted: 02/07/2019 at 5:14pm
All -

As you know, I am no longer with USATT. However, I can assure you (unless something has changed in the last month) the algorithms have not changed and are not secret in any way shape or form EXCEPT for the change, which we told the membership about in 2016, to address the "all wins" situation where under the old algorithms you could actually be penalized for winning all of your matches instead of strategically losing one. That is the only thing that I am aware of that has changed.

If someone wants to send me the specifics of this particularly situation, I would be happy to look at it.

Thanks
JOOLA Nobilis / SP (FH)/ Rhyzm Tech (BH)
Back to Top
Gordy View Drop Down
Silver Member
Silver Member


Joined: 12/04/2011
Location: United States
Status: Offline
Points: 648
Post Options Post Options   Thanks (0) Thanks(0)   Quote Gordy Quote  Post ReplyReply Direct Link To This Post Posted: 02/07/2019 at 5:15pm
@Push - "Well, to be exact, it's the 3rd and final pass.. Pass 2 determines adjustments and Pass 3 determines the final rating.. Yes, the adjusted rating of your opponents is being used for the 3rd pass and the pre-adjustment rating for the first pass as also for pass 2 which determines adjustments" is only partially right. There are actually 4 passes - Pass 3 actually brings in previously unrated players.
JOOLA Nobilis / SP (FH)/ Rhyzm Tech (BH)
Back to Top
Gordy View Drop Down
Silver Member
Silver Member


Joined: 12/04/2011
Location: United States
Status: Offline
Points: 648
Post Options Post Options   Thanks (0) Thanks(0)   Quote Gordy Quote  Post ReplyReply Direct Link To This Post Posted: 02/07/2019 at 5:29pm
OK, no need to send me the particulars, I was able to figure out who it was.

Looks like he may have a point...

Pre-Tournament Rating: 2072
Win vs. 2339 = 50 points
Win vs. 2186 = 25 points
Loss vs. 2545 (pre tournament rating) = 0
Net gain = 75 points

Explanation says "The second tier is used only for those rated players who have experienced a rating change of at least 75 for a particular tournament. If the player has wins and losses, the Adjusted Rating is derived by taking the average of the player’s Pre-Tournament Rating, and the average of the player’s best win and worst loss."

Therefore the adjustment should have been the average of 2072 and 2442, or 2257.

Then applying the 2257 to the PASS 4 points calculation, the final rating should have been:

Win vs. 2339 = (-82 upset) = +16
Win vs. 2186 = (71 point favorite) = +5
Loss vs. 2545 = 0

2257 + 16 + 5 = 2278 should be the final rating.

Of course I can't look "behind the scenes," but the data shows this is what it should be. It is possible I am missing something...
JOOLA Nobilis / SP (FH)/ Rhyzm Tech (BH)
Back to Top
Gordy View Drop Down
Silver Member
Silver Member


Joined: 12/04/2011
Location: United States
Status: Offline
Points: 648
Post Options Post Options   Thanks (0) Thanks(0)   Quote Gordy Quote  Post ReplyReply Direct Link To This Post Posted: 02/07/2019 at 5:32pm
Also, to all...when you are looking at adjustments in general, make sure to look at the PRE TOURNAMENT RATINGS for all players. It is common error we come across where someone gains, say +55, in the PASS4 but don't look at the pre-tournament ratings...
JOOLA Nobilis / SP (FH)/ Rhyzm Tech (BH)
Back to Top
tom View Drop Down
Gold Member
Gold Member


Joined: 11/18/2013
Location: canada
Status: Offline
Points: 1477
Post Options Post Options   Thanks (0) Thanks(0)   Quote tom Quote  Post ReplyReply Direct Link To This Post Posted: 02/07/2019 at 6:04pm
finally an explanation that I can follow.  I think the OP said the loss was against a 2550 ( vs. 2545).  Gordy,  could you explain if he had lost to a 2800 instead of 2550( or 2445) his adjustment  would have been (2072+(2800+2339)/2)/2= 2320 or 2321(depending on the rounding).  By losing to a higher rated player his initial adjusted rating would have been 2320 instead of 2257???

Edited by tom - 02/07/2019 at 6:06pm
Back to Top
Gordy View Drop Down
Silver Member
Silver Member


Joined: 12/04/2011
Location: United States
Status: Offline
Points: 648
Post Options Post Options   Thanks (0) Thanks(0)   Quote Gordy Quote  Post ReplyReply Direct Link To This Post Posted: 02/07/2019 at 6:17pm
@Tom - the OP thought he lost to a 2550 because that was his adjusted rating. His pre-tournament rating (the 2550 player) was 2545.

I am not going to like the answer, Tom...but yes, you are correct. Had his only loss been against a 2801 rated player (assuming the 2801 was his pre-tournament rating), his adjusted rating would have been:

2072 + 2570 (the average of 2339 and 2801) divided by 2. In other words, his adjusted rating (pre-pass 4) would have been 2321. (I used 2801 so there was no rounding)


Edited by Gordy - 02/07/2019 at 6:18pm
JOOLA Nobilis / SP (FH)/ Rhyzm Tech (BH)
Back to Top
smackman View Drop Down
Premier Member
Premier Member
Avatar

Joined: 07/20/2009
Location: New Zealand
Status: Offline
Points: 3124
Post Options Post Options   Thanks (0) Thanks(0)   Quote smackman Quote  Post ReplyReply Direct Link To This Post Posted: 02/07/2019 at 6:18pm
I have it all worked out , just by looking at the answers above and below
your new rating should be
2147
or 2199
or 2200
or 2257
or 2258
or 2278
or 2280
or 2300+-
or 2312
or 2320
or 2400
or 2500
you Americans have a complicated system we would just get the 75 points here


Edited by smackman - 02/08/2019 at 6:20pm
Ulmo Duality,tibhar Aurus Prime Dr N Pistal Black
NZ table tennis selector, third in the World (plate Doubles)I'm Listed on the ITTF website,
Back to Top
pongfugrasshopper View Drop Down
Gold Member
Gold Member
Avatar

Joined: 03/22/2015
Location: USA
Status: Offline
Points: 1029
Post Options Post Options   Thanks (0) Thanks(0)   Quote pongfugrasshopper Quote  Post ReplyReply Direct Link To This Post Posted: 02/07/2019 at 6:59pm
Originally posted by Gordy Gordy wrote:

OK, no need to send me the particulars, I was able to figure out who it was.

Looks like he may have a point...

Pre-Tournament Rating: 2072
Win vs. 2339 = 50 points
Win vs. 2186 = 25 points
Loss vs. 2545 (pre tournament rating) = 0
Net gain = 75 points

Explanation says "The second tier is used only for those rated players who have experienced a rating change of at least 75 for a particular tournament. If the player has wins and losses, the Adjusted Rating is derived by taking the average of the player’s Pre-Tournament Rating, and the average of the player’s best win and worst loss."

Therefore the adjustment should have been the average of 2072 and 2442, or 2257.

Then applying the 2257 to the PASS 4 points calculation, the final rating should have been:

Win vs. 2339 = (-82 upset) = +16
Win vs. 2186 = (71 point favorite) = +5
Loss vs. 2545 = 0

2257 + 16 + 5 = 2278 should be the final rating.

Of course I can't look "behind the scenes," but the data shows this is what it should be. It is possible I am missing something...

Miscalculations like this shouldn't be happening years after the system has been put in place wouldn't you agree?  I suspect the OP will eventually get it corrected.  

It's now been corrected, but a month ago, pgpg posted about how Uezu Koji got a weird rating after the US Open.  Just doesn't give us the warm and fuzzies when stuff like this continues to happen.
Back to Top
pgpg View Drop Down
Gold Member
Gold Member
Avatar

Joined: 11/18/2013
Location: United States
Status: Offline
Points: 1089
Post Options Post Options   Thanks (0) Thanks(0)   Quote pgpg Quote  Post ReplyReply Direct Link To This Post Posted: 02/07/2019 at 7:56pm
Originally posted by pongfugrasshopper pongfugrasshopper wrote:

...
It's now been corrected, but a month ago, pgpg posted about how Uezu Koji got a weird rating after the US Open.  Just doesn't give us the warm and fuzzies when stuff like this continues to happen.

I believe Larry Hodges had to intervene to make it happen...
USATT: ~1900
Donic Defplay - AK47 blue - Dtecs OX
Back to Top
 Post Reply Post Reply Page  <1234 5>
  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 0.094 seconds.

Become a Fan on Facebook Follow us on Twitter Web Wiz News
About MyTableTennis.NET | Forum Help | Disclaimer

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

Copyright ©2003-2019 Alex Table Tennis Ltd. All rights reserved.