- Blind mode tutorial
lichess.org
Donate

Rating System Broken.

#111

@n321

"Please do not start personal accusations."

That's quite rich coming from someone who declared the rating system "plain rubbish" in another thread and "glorified due to its intransparency" here.

What is the basis for your accusations? Have you tested Glicko-2? No! Have you tested it in comparison to an ELO implentation? No, of course not. Did you read Glickman's description of Glicko-2 and assessed the reasons Glickman gave for his changes of the algorithm over time? I guess no, otherwise you would have spared us your rantings about "intransparency".

And how was my comment a personal accusation? You stated in #97 quite clearly what the reason for your unhappiness with the rating system is: You wanted more rating points awarded for your so-called winning streak. Even more so, you told us you "needed" them. If that's not fixation then what is?

@n321 "Please do not start personal accusations." That's quite rich coming from someone who declared the rating system "plain rubbish" in another thread and "glorified due to its intransparency" here. What is the basis for your accusations? Have you tested Glicko-2? No! Have you tested it in comparison to an ELO implentation? No, of course not. Did you read Glickman's description of Glicko-2 and assessed the reasons Glickman gave for his changes of the algorithm over time? I guess no, otherwise you would have spared us your rantings about "intransparency". And how was my comment a personal accusation? You stated in #97 quite clearly what the reason for your unhappiness with the rating system is: You wanted more rating points awarded for your so-called winning streak. Even more so, you told us you "needed" them. If that's not fixation then what is?
#112

@n321 obviously you lack sufficient backround in maths to understand this BUT this stament
"@BlackSalt

Better not to mix data from one to another system, we will get quite weird results and go nowhere with conclusions.
"

Nope thet are very very similar systems. Both assume same expected outcomes for same point difference. so if you take Glicko generated numbers and feed them performance number evaluation you get perfectly valid results. Performance number is not even part of elo system and definately is by no means Elo specific. And Bradley-Terry style performance indicator will do. Dont bother figting the rating system . THere are better ones but this one works just fine. Win more and lose less. and your ratingwill go up

@n321 obviously you lack sufficient backround in maths to understand this BUT this stament "@BlackSalt Better not to mix data from one to another system, we will get quite weird results and go nowhere with conclusions. " Nope thet are very very similar systems. Both assume same expected outcomes for same point difference. so if you take Glicko generated numbers and feed them performance number evaluation you get perfectly valid results. Performance number is not even part of elo system and definately is by no means Elo specific. And Bradley-Terry style performance indicator will do. Dont bother figting the rating system . THere are better ones but this one works just fine. Win more and lose less. and your ratingwill go up
#113

I don't disagree that players would likely be better off improving than attempting to scrutinize the rating system, but both are possible.

I don't disagree that players would likely be better off improving than attempting to scrutinize the rating system, but both are possible.
#114

@HellevatorOperator "Why aren't you focused on having fun?"

I do actaually having fun here. And all this is about proving my point that: "When a rat fall into the trap after dozen of lost games at beginning losing 500 or even all 700, cannot recover ever with this implemented rating system here".

But the most of the people here seem to missing all that, current mood for play etc, but only looking at current rating in blitz only I'm playing here (3+0 to 5+4). For these here is quite simple question: How that is possible that I have almost 2200 in puzzles, as rating systems is generally the same? (actually it is rhetorical question).

Playing few games a day show quite a bit difference in W/L ratio and actually consistency and quality in blitz games.

@HellevatorOperator "Why aren't you focused on having fun?" I do actaually having fun here. And all this is about proving my point that: "When a rat fall into the trap after dozen of lost games at beginning losing 500 or even all 700, cannot recover ever with this implemented rating system here". But the most of the people here seem to missing all that, current mood for play etc, but only looking at current rating in blitz only I'm playing here (3+0 to 5+4). For these here is quite simple question: How that is possible that I have almost 2200 in puzzles, as rating systems is generally the same? (actually it is rhetorical question). Playing few games a day show quite a bit difference in W/L ratio and actually consistency and quality in blitz games.
#115

There is no trap. If you can not recover you are not good enough.

How come you have almost 2200 in puzzles and not in blitz? Because puzzles are not timed and puzzles only test tactics and not time control, strategy, positioning, opening understanding and playing under pressure. There is lots that puzzles don't test and most people have a higher puzzle rating than actual rating?

What about all the people above you? Are they stuck in a trap? If the rating system is broken, how come everyone is not stuck in a trap? Or just special players?

There is no trap. If you can not recover you are not good enough. How come you have almost 2200 in puzzles and not in blitz? Because puzzles are not timed and puzzles only test tactics and not time control, strategy, positioning, opening understanding and playing under pressure. There is lots that puzzles don't test and most people have a higher puzzle rating than actual rating? What about all the people above you? Are they stuck in a trap? If the rating system is broken, how come everyone is not stuck in a trap? Or just special players?
#116

@HellevatorOperator

@Toadofsky gave enough info to check all regarding G2 implementation here in #105 and that is enough for futher research.

Nothing to comment regarding personal accusations and insinuations. My intetntion here is to dissect G2 implemented here in Lichess, not to waste time with people who see only current rating and nothing else. Chess is a game as any - if it is not your day and force yourself to play several dozen of games in a row (50 in this case after registration), it is quite logical that you will lose.

@HellevatorOperator @Toadofsky gave enough info to check all regarding G2 implementation here in #105 and that is enough for futher research. Nothing to comment regarding personal accusations and insinuations. My intetntion here is to dissect G2 implemented here in Lichess, not to waste time with people who see only current rating and nothing else. Chess is a game as any - if it is not your day and force yourself to play several dozen of games in a row (50 in this case after registration), it is quite logical that you will lose.
#117

I win aganist 2750 at bullet (old account) but just gain 8 points
because rating deviation

I win aganist 2750 at bullet (old account) but just gain 8 points because rating deviation
#120

The problem I see with the rating system right now is that there seems to be a cap on how many points you gain per game no matter how much stronger your opponent is, which seems to be about 12. This means that there is no quick easy fix for losing the first few games on Lichess since they are weighted so heavily. I think this is quite unfair to those who lost their first few games but improved, and now cannot get out of the trap. This is the only aspect of the rating system in need to fixing: the fact that it does not account for improvements and early mistakes, 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?

The problem I see with the rating system right now is that there seems to be a cap on how many points you gain per game no matter how much stronger your opponent is, which seems to be about 12. This means that there is no quick easy fix for losing the first few games on Lichess since they are weighted so heavily. I think this is quite unfair to those who lost their first few games but improved, and now cannot get out of the trap. This is the only aspect of the rating system in need to fixing: the fact that it does not account for improvements and early mistakes, 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?

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