I need to speed up an algorithm that takes as input the number of chips of several players in a tournament. Their chips can be seen as a normally distributed curve.
Since my algorithm can handle a maximum of 10 inputs and its purpose is to calculate each player probability of ending the tournament in a determined place I thought to divide the gaussian curve in 10 «ranges» and put them as input of the algorithm.
Since I'm not a mathematician I would really appreciate to hear your opinion of this method, are the outputs trustworthy? Or they'll have a significant level of error? Should my 10 ranges be divided equally or should I take into consideration the fact that there are more players in the central ranges and divide them proportionally to that?
Thanx in advance Giorgio