This post is for Pegasus, he has requested information about how to compute weights based on assumptions of a normally distributed factor.
The process is very simple, first take the factor that you want to use and analyze the distribution for example this one:
That distribution is mostly centered around 5 with a couple a big outliers so the weights should reflect that.
1. Compute the mean and standard deviation:
mean = 5, sd = 2.318405
2. Rank each value based on the probability density they have, here is the list of densities for each value:
0.03884446 0.15679108 0.16812061 0.17207621 0.16812061 0.16812061 0.16812061 0.17207621 0.01681642
3 Finally normalize using max density to get the weights (divided everything by the max on the previous list):
0.22573984 0.91117231 0.97701252 1.00000000 0.97701252 0.97701252 0.97701252 1.00000000 0.09772661
Those are the weights to be used. Please let me know if you have any other questions.