|
BGonline.org Forums
error rate tournament checker play
Posted By: Maik Stiebler In Response To: error rate tournament checker play (bob koca)
Date: Saturday, 9 June 2007, at 7:07 a.m.
Instead of just discounting forced moves, discount moves in which every play gives the same equity.
But in the situation that you have constructed, that would not change anything, as playing an ace 2-1 in the next move would usually be a blunder, right?
More generally, I agree with Matt that total error is the natural thing to look at (it is debatable if normalization to EMG or similar should be used or just raw MWC). Snowie style ER is much closer to that natural metric than gnubg style. I know that there are many who would not agree at all, but I honestly cannot see a single good reason why gnubg style ER, modified or not, should ever be used. Ok, there might be one reason, that for most players gnubg style ERs have less variance, but someone has to show me statistical evidence before I believe it.
Isn't it true that gnubg style ER unnaturally rewards e.g. blitzish plays and late doubling, just because they tend to increase the number of your decisions relative to your opponent's?
|
BGonline.org Forums is maintained by Stick with WebBBS 5.12.