Rating System Broken.

#120
"...no matter how much stronger your opponent is"

Supposing you beat Carlsen, your rating would spike. But also 12 points per blitz game (this isn't a cap) is quite a bit; I have frequently gained or lost 50-100 points per hour.

You are entitled to opinions about fairness (an opinion some people share) however we are not walking the path USCF unfortunately walked when writing their 200-page rulebook (which to their credit, they are trying to improve but any sort of complex change is difficult). We are especially not changing rules based upon fabrications:
"...instead, fixing rating at a steady level for eternity"

"I have personally never seen anyone return to a provisional rating after a few games, perhaps this needs to be easier to get to?"

I wrote a patch that ensured player ratings would return to provisional after a year of inactivity instead of slightly longer (~400 days if I recall), however it was rejected for a lack of testing on my part (since I lack a good way to test all the effects that might have). There might be some merit to your idea although without adequate testing we cannot know. People have all kinds of ideas about how to change the rating system but very few people are able and willing to put in the difficult research effort to prove that changes would be positive (or at least prove they would not be negative).

#120 "...no matter how much stronger your opponent is" Supposing you beat Carlsen, your rating would spike. But also 12 points per blitz game (this isn't a cap) is quite a bit; I have frequently gained or lost 50-100 points per hour. You are entitled to opinions about fairness (an opinion some people share) however we are not walking the path USCF unfortunately walked when writing their 200-page rulebook (which to their credit, they are trying to improve but any sort of complex change is difficult). We are especially not changing rules based upon fabrications: "...instead, fixing rating at a steady level for eternity" "I have personally never seen anyone return to a provisional rating after a few games, perhaps this needs to be easier to get to?" I wrote a patch that ensured player ratings would return to provisional after a year of inactivity instead of slightly longer (~400 days if I recall), however it was rejected for a lack of testing on my part (since I lack a good way to test all the effects that might have). There might be some merit to your idea although without adequate testing we cannot know. People have all kinds of ideas about how to change the rating system but very few people are able and willing to put in the difficult research effort to prove that changes would be positive (or at least prove they would not be negative).

Katzenschinken

#122

@Toadofsky

"I have frequently gained or lost 50-100 points per hour."

Exactly. There is no rating hell in which you are stuck forever when you mess up the first games.

Plus I don't think that anyone could mess up so badly with his first games that he lands at 1300 points when in reality his skill is good enough for 2000 points. If you have 1300 points then you more or less deserve to be there. And if you get better and play enough games this will be reflected by your rating soon enough, just like for anybody else.

@Toadofsky "I have frequently gained or lost 50-100 points per hour." Exactly. There is no rating hell in which you are stuck forever when you mess up the first games. Plus I don't think that anyone could mess up so badly with his first games that he lands at 1300 points when in reality his skill is good enough for 2000 points. If you have 1300 points then you more or less deserve to be there. And if you get better and play enough games this will be reflected by your rating soon enough, just like for anybody else.

MarkusNemetz

#123

Just a question

MarkusNemetz (2321) +1
bbeg (2319) ±0

in a draw in a correspondence game (

https://lichess.org/02J7usz9X0m0

)

How is this possible?

Just a question MarkusNemetz (2321) +1 bbeg (2319) ±0 in a draw in a correspondence game (https://lichess.org/02J7usz9X0m0) How is this possible?

sheckley666

#124

Rounding + maybe different rating deviation

drmrboss

#125

The best way to get out of 1300 rating pool is
" dont play vs 1300". That is why 2300 seek vs 2300 and maintain their ratings.

If you lose the first 10 games, then you will be 1300 and then you keep playing vs 1300 .

It will be pretty hard for you to get out of that pool.

The best way to get out of 1300 rating pool is " dont play vs 1300". That is why 2300 seek vs 2300 and maintain their ratings. If you lose the first 10 games, then you will be 1300 and then you keep playing vs 1300 . It will be pretty hard for you to get out of that pool.

petri999

#126

No it is not. I have had losing streaks to drop from 1500+ to well below 1300 and then back.

If you are really a 1500 player and you play 1300 player you lose about 1/4 games to a 1300 player in four games gaining about 15-18 points and losing 5 with total gain of 10 points. So getting to around 1400 would take about 40 games. At 1400 with true strength of 1500 you woudl still win about 65% games so progress would slow down not by much though.

And if the rating system would be faster and would still give reward for each game separately then it would also make your losing streak go down faster. And climbing back so much harder

Elo like system (Glicko is one of them) it is a known system that playing weaker players might hurt your chances. But that is hard to fix at least according to Jeff Sonas https://en.chessbase.com/post/what-s-wrong-with-the-elo-system
Ans frankly that is least of problems in Elo-system

No it is not. I have had losing streaks to drop from 1500+ to well below 1300 and then back. If you are really a 1500 player and you play 1300 player you lose about 1/4 games to a 1300 player in four games gaining about 15-18 points and losing 5 with total gain of 10 points. So getting to around 1400 would take about 40 games. At 1400 with true strength of 1500 you woudl still win about 65% games so progress would slow down not by much though. And if the rating system would be faster and would still give reward for each game separately then it would also make your losing streak go down faster. And climbing back so much harder Elo like system (Glicko is one of them) it is a known system that playing weaker players might hurt your chances. But that is hard to fix at least according to Jeff Sonas https://en.chessbase.com/post/what-s-wrong-with-the-elo-system Ans frankly that is least of problems in Elo-system

Morozov edited

#127

@MarkusNemetz it is entirely possible. Rating has decimal values. 2321,84 vs 2319,52 for example. 0,3 points lost/gained changes the displayed value for one player but not for the other one.

Edit: Rating is not rounded up or down. And "Elo Hell" was made up by people who like to blame others for their failure.

@MarkusNemetz it is entirely possible. Rating has decimal values. 2321,84 vs 2319,52 for example. 0,3 points lost/gained changes the displayed value for one player but not for the other one. Edit: Rating is not rounded up or down. And "Elo Hell" was made up by people who like to blame others for their failure.

vio7

#128

@MarkusNemetz
in correspondence games i believe it takes the rating
for calculation when the game is finished and not the rating when the game was started.

@MarkusNemetz in correspondence games i believe it takes the rating for calculation when the game is finished and not the rating when the game was started.

ShootingStar34 edited

#129

Oh gosh!!!! Once me and my opponent's scores were not far apart at all and he had more points than me and they made me lose 58 points, an opponent, HIGHER THAN ME, I had 1720 and he had 1828. It was an RK game, 58 points lost? For what?? Nonsensical. I played in an Antichess tournament and I beat a 2180 and I had 1470, I only got 48+, shouldn't I get like over 100 points?

david9426

edited

#130

wait is that person me shootingstar

This topic has been archived and can no longer be replied to.