[ QUOTE ]
Curious, that's very different from the stats I came up with. Obviously due to the fact of 'conditioning' against the upper bound of 25%. I'm in agreement with the reasoning, but how exactly does that work?
This puts his most-likely ROI around 17-18%.
[/ QUOTE ]
I'm no statistician (more of an applied maths person) so I might have fudged some things. I took the output from the ROI simulator (16/13/14.75 dist 35% ROI) and fit it to a normal distribution. I then combined this with another modified normal distribution with mean -9.09 (rake) which went to zero at 25, to represent the distribution of players ROIs. The standard deviation of players ROIs was chosen fairly subjectively of course, heres a more optimistic output, it assumes ~1/3 of players beat the rake: