6 edits
Originally posted by CenterNutmaybe you should also have a close look at archbold 😉
he from the UK.. he got my support 😵
1 loss: Game 121723 only a timeout,
and
also beats webreaper rated 2123 Game 445513
and
beats The PO currently rated 2280 Game 340701
unfortunately no strong players seem to be playing archbold at the moment ... 🙄
Originally posted by flexmoreI just did a comparison between IronMan31 and Northern Lad, calculating their performance over each of their last 150 games. I took some liberties in preparing the data (it was far simpler to use their opponents current rating than the one used at game completion).
what would northern lad's rating be if he repeated all the games he has played so far with the same results ... then repeated it all again and again, again?
would he be rated as high as ironman?
can this calculation be easily made by anyone?
what about trying the same with weyerstrass and akizy!
it would be interesting to see this calculation nex ...[text shortened]... yer in the top 40 players: ... what would their rating be if they repeated their last 300 games?
Northern Lad's 150 game performance is 2350
IronMan31's 150 game performance is 2496
Computing performance does not take into account *your* rating, only the result and the rating of your opponent.
Originally posted by flexmoreWOWWWWWWW good games
maybe you should also have a close look at archbold 😉
1 loss: Game 121723 only a timeout,
and
also beats webreaper rated 2123 Game 445513
and
beats The PO currently rated 2280 Game 340701
unfortunately no strong players seem to be playing archbold at the moment ... 🙄
1 edit
Originally posted by RookRAKhow about archbold if you get rid of the timeout?
I just did a comparison between IronMan31 and Northern Lad, calculating their performance over each of their last 150 games. I took some liberties in preparing the data (it was far simpler to use their opponents current rating than the one ...[text shortened]... t *your* rating, only the result and the rating of your opponent.
- and akizy ?!
Originally posted by RookRAKwell done thankyou for the calculations!
I just did a comparison between IronMan31 and Northern Lad, calculating their performance over each of their last 150 games. I took some liberties in preparing the data (it was far simpler to use their opponents current rating than the one used at game completion).
Northern Lad's 150 game performance is 2350
IronMan31's 150 game performance is 2496
...[text shortened]... ance does not take into account *your* rating, only the result and the rating of your opponent.
i am just wondering about your exact method ...
is it:
rating = ((Wav+ K)*Nw + Dav*Nd + (Lav-K)*Nl)/(Nw + Nd + Nl)
where:
Wav = average rating of players won against,
Dav = average rating of players drawn with,
Lav = average rating of players lost to,
and
Nw = number of players won against,
Nd = number of players drawn with,
Nl = number of players lost to.
and
K = constant >= 0.
?
Originally posted by flexmoreRather than use average ratings, I actually do the calculation for every game. Using an iterative method, I converge on the rating which produces the lowest total error.
i am just wondering about your exact method ...
is it:
rating = ((Wav+ K)*Nw + Dav*Nd + (Lav-K)*Nl)/(Nw + Nd + Nl)
?
I.e. For each iteration; for each game; I calculate:
error = actual performance - expected performance
I then sum the errors for all games, using that to generate the next "guess" at rating
There is no K-factor needed for calculating performance. K-factor only comes into play when you are actually modifying a rating.