the point system we all live with, a fist winn and 2 = 1759 10 points for workes in loss to a 3 or 7 games manny points, its only the starters getting the big swing louse to a good player and stay the same points,
2 games 1759 or defending on opponent, you louse 300 points to a 1112
it happens or like that, sorry my english
it happens or like that, sorry my english
Merci Jacques.
Why is that jim?
Aren't this numbers supposed to balance the difference of ability so that, after a long serie of game, the 2 players have roughly the same elo as when they start?
It might not be totally accurate as glicko is a bit tricky. I assume it's still fairly true.
Why is that jim?
Aren't this numbers supposed to balance the difference of ability so that, after a long serie of game, the 2 players have roughly the same elo as when they start?
It might not be totally accurate as glicko is a bit tricky. I assume it's still fairly true.
Sorry Bramaputra, i don't see where you are going.
We are talking about players with a high RD here.
We are talking about players with a high RD here.
a 1500 player start is like
the other with manny games a 1500 gett hig fast and down fast to,we dont do,who play manny games,
the other with manny games a 1500 gett hig fast and down fast to,we dont do,who play manny games,
"Aren't this numbers supposed to balance the difference of ability so that, after a long serie of game, the 2 players have roughly the same elo as when they start?"
Why would that make sense?
Anyway, the reason it was incorrect is because rating changes depend on both player's rating deviation, not just the displayed rating
This means that you could for example gain 5 points against one 1700 but 20 points against another 1700 even when your rating was the same for both. So, trying to translate these numbers into probabilities of winning is a meaningless endeavour
Why would that make sense?
Anyway, the reason it was incorrect is because rating changes depend on both player's rating deviation, not just the displayed rating
This means that you could for example gain 5 points against one 1700 but 20 points against another 1700 even when your rating was the same for both. So, trying to translate these numbers into probabilities of winning is a meaningless endeavour
Why does nobody want to make it easy for me?
We are talking about players with a high RD here.
In french, we have this saying : "enculer les mouches" a lot of trouble for very little reward.
In a regular elo system it's totally relevant..
Mr Elo came up with those coefficients which are a pure convention and millions of games later, players' ratings were distributed according to what approaches a normal law. Therefore the gain/loss ratio represents well enough those odds.
And Glicko.2 is close enough to me. We all know tat it's not linear, it's multidimensional. Thank you for reminding us. But we all rely upon it : it's close enough.
Maybe you believe there is out there, in the ether of that platonic world of Ideas, the true law of probability for that problem swirling amongst the tetrahedrons and the concepts of liberty and monotheism. I don't.
It's no intellectually satisfactory to replace a model by another one, glicko by elo...i understand. But i'd like to point out that any mathematical construction is only true relatively to itself. That any mathematical system whether it is physical, biological, sociological... is never true but only consistent with the observations.
If you believe that because i spent 2 weeks sunbathing in the south Atlantic with no internet and because my RD went too low i became another player, let me disagree.
OK i'm pissed, did anyone notice?
Bad idea, my fault. Sorry for wasting the time of you all.
We are talking about players with a high RD here.
In french, we have this saying : "enculer les mouches" a lot of trouble for very little reward.
In a regular elo system it's totally relevant..
Mr Elo came up with those coefficients which are a pure convention and millions of games later, players' ratings were distributed according to what approaches a normal law. Therefore the gain/loss ratio represents well enough those odds.
And Glicko.2 is close enough to me. We all know tat it's not linear, it's multidimensional. Thank you for reminding us. But we all rely upon it : it's close enough.
Maybe you believe there is out there, in the ether of that platonic world of Ideas, the true law of probability for that problem swirling amongst the tetrahedrons and the concepts of liberty and monotheism. I don't.
It's no intellectually satisfactory to replace a model by another one, glicko by elo...i understand. But i'd like to point out that any mathematical construction is only true relatively to itself. That any mathematical system whether it is physical, biological, sociological... is never true but only consistent with the observations.
If you believe that because i spent 2 weeks sunbathing in the south Atlantic with no internet and because my RD went too low i became another player, let me disagree.
OK i'm pissed, did anyone notice?
Bad idea, my fault. Sorry for wasting the time of you all.
Your arguments are completely incoherent. For your sake, I will assume that meaning is being lost in translation
berserk your idea (result vs expected result) is very good that 's why you have so much poor opposition :
- there are many cases where this is useless (like alarms, salt on the table, ...)
- on your supposed aggressivity (missunderstood humour ?)
- on wrong numbers (Pi is NOT 3,14 )
- there are many cases where this is useless (like alarms, salt on the table, ...)
- on your supposed aggressivity (missunderstood humour ?)
- on wrong numbers (Pi is NOT 3,14 )
Very good suggestion, I hope it comes to fruition! I'll elaborate further in a later post =)
This topic has been archived and can no longer be replied to.