Chris;

> yes, Roger, that is what you need to say for someone to know what you were doing.

Good.

> What I say below applies partly to a post on this page, but maybe more to evaluations that you might do professionally...for SCEC for example.

Doesn't look like that will happen

> The one thing that you may not have explained well enough is the z-binomial technique....but since you do not use it in the case of Don, it is OK to not explain it.

It's a standard test for when you have a lot of predictions with the same probability.

> You did describe pretty well the method of Alan Jones, but, if that method is published, you would want to reference it, in case someone wanted to know more.

It isn't. He and his brother devised it at my urging. I'm trusting him that it works.

> I did not 100% follow everything but that is my problem this time, not yours...I don't understand math very well (despite 8 college courses in it), and have limited exposure to statistics.

Me too actually. I struggled thru calculus and said "That's enough!". Never have needed to use it either.

> But...you might want to explain what a standard deviation of 4.299 means. What would it be from a random series of predictions?

It's 4.299 standard deviations on a normal curve. Refers to the area under the curve. My table only goes to 3.70 and at that point only 0.0001 remains to the right.

Or to put it simply, d**n little chance of this happening by chance.

> As for the comment of Don...that confuses me. He seems to be talking about something different. You are evaluating predictions that have been made, and I don't see what future predictions or technique for the prediction have to do with this.

Nor do I. You can evaluate past success as I do or you can evaluate the likelihood that a particular method will succeed.

Personally, I don't care what method is used, I just want to know how well they are doing.

Roger