The speed issue is on my end. Because I wrote the analysis code to be generalized, it computes the Jones score (and EV and VAR) for every prediction. In this case, the Jones score never changes since the magnitude never changes. The EV and VAR only has two possibilities based on hit or miss. The upshot is I'm wasting CPU time calculating these things over and over.

A highly optimized program could easily be made for this specific test where only the two possibilities are calculated once - hit or miss. Then, make the random list and find how many hits. Then multiply the hit values by the number of hits, multiply the miss values by the number of misses, then add the two together to get the total sums for each set of predictions. This method would probably only take a minute or two.

I'll let you know when I get it done.

Brian