|
Post by steadyeddie on Jul 8, 2015 14:30:22 GMT -8
I play a quite a bit of chess and I'm familiar with Elo rankings. My chess ranking is fairly stable. It doesn't matter if I play strong or weak players. If I play really strong players and get 10% that's OK.
But on Tiltyard I find if I play strong players my players rating dives. I mean, I think 10-20% against Sancho is respectable, while 70% against Coursera newbies shouldn't be enough. But the ratings changes don't reflect that.
Let's say I'm ~50 on tiltyard, and Sancho is ~200. What am I expecting to score to "break even"? Does Agon dificulty matter? Surely a game of tictactoe should be discounted as it is almost trivially drawn.
|
|
|
Post by Steve Draper on Jul 8, 2015 18:10:13 GMT -8
It's a population stability issue I think. The population on Tiltyard is far too small to iron out statistical fluctuations so as significant players come and go on timescales of a day or so it can significantly move the 'average skill' of the population, which causes everyone to migrate (there is no absolute fixed point).
Sam can provide more details on how it is all calculated, but ELO is precisely that (exactly the same definition as in chess etc.), and AGON also takes account of (statistically apparent) game and role differences, but is broadly a similar system.
|
|