Rating System Broken.

Hi guys,
I just beat a 2200 bullet player in this game (bullet game): lichess.org/F5kRfNQM/white#51

I only got 11 points when he was 500 points higher than me. I have had many other games, beating overrated players and only got a few points. I think you should get more rating points for defeats like this.

Another thing is that if you were to play to games against someone who is same rating and lose twice then you would lose around 11 points which is what you gain for beating someone who is 500 points higher.

I know that lichess uses a particular rating system but i don't think that this really works because of the thoughts above.

11 is quite a big number!

Seriously: Your rating deviation in bullet is 45. What do you expect?

@Fischer_and_Chips88

Similar here, but 400 points difference:

@Fischer_and_Chips88 "Another thing is that if you were to play to games against someone who is same rating and lose twice then you would lose around 11 points which is what you gain for beating someone who is 500 points higher."

AFAIK, this is also true for the FIDE Elo system.

Stop whining, theres nothing wrong with the rating system. You arent going to get a lot of points for a single game. To get a higher rating you need to get better, fluke wins now and then are irrelevant to your strength.

@sheckley666 "AFAIK, this is also true for the FIDE Elo system."

Let check: ratings.fide.com/calculator_rtd.phtml

"Difference of more than 400 is counted as 400.
Rating Change is: +18.4"

Rating = 1235
Rc = 1645
W = 1
K val = 20

And what does it say for losing twice against a player of same rating? I guess -20.

For losing per game from 400+ higher rated player is just -1.6, for draw +8.4.

But here, on Lichess, rating is quite chaotic.

That's not quite what I had asked. But I tested it myself, and I was right :)

The basic idea is the same for Elo and Glicko-2 used by Lichess. The value depends on the difference between the actual result and the expected result. The expected result for a game against an equal rated player is 0.5, so for a loss the difference is -0.5
The expected result against a much better player is close to zero, so for a win the difference is close to 1.

And therefore you lose for two losses against an equal rated opponent roughly the amaount of points, that you can win against a much higher rated opponent.

The difference between FIDE and Lichess is mainly the development coefficient K, which is much more sophisticated here on Lichess.

Glicko-2 implemented here is hardly "sophisticated". It is just more complicated and restrictive, especially when start.

I do not know exactly what rating system chess.com use, but it starts from 400, not 1500.

IIRC, the chess.com offer possibility to clean rating and start all over from beginning, but here that possibility doesn't exists.