Ara;

Once again a post which only exposes your ignorance.

But I'll try to reduce it somewhat.

First off, there are two methods to handle predictions like Don's and Shan's which are numerous but have different probabilities.

The first, devised by NetSurfr, involves selecting a random number for each prediction. That number is compared to the real prediction and if it is smaller than the real probability, a hit is counted. Repeat for each prediction and store the number of hits in a counting array.

Do this 10,000 times and you will get a normal distribution. Then see where the hit count for the real predictions fall and you can determine the significance from that.

The second method comes from Dr Alan Jones. It's a simpler method involving only a few equations but I'd have to post the program for you to understand it.

Roger