Quote:
Originally Posted by dave_w11
The results of HEM agree with Juks program very well so I think the calculations themselves are working accurately, but there could be some kind of bias introduced that we don't understand, perhaps linked in to the inaccuracies of ICM for example.
After using my own program quite alot I'm pretty certain that there is some degree of correlation between your $EV_luck and you actual winnings. I often see times when I am super-lucky or super-unlucky and it seems to have a scaling effect on my adjusted winrate.
The times where we get super-lucky and the gradient scales upward then I think we can put that down to letting us (ie: a winning player) get further into the SNG and thus be able to better take advantage of our skill-edge and general "donk equity". The same goes for when we get super-unlucky and the gradient is scaled down because we don't get the chance to take advantage of our skill-edge and miss out on the future "donk equity".
I think there is more to it than this (ie: sometimes, paradoxically, you can get super-super-lucky and it appears to scale the gradient downwards, etc), but I think this is basically it and the problem lies with the limitations of the ICM function itself (ie: the "equal skill" assumption).
Juk