Another point to consider is that the parameters are not linear. Magnitude is exponential, area of a circle is not linear.

I feel it is up to the predictor to state his limits, variable or not and up to the evaluator to decide on the "worth" of any possible hit. A track record is an absolute must in determining reliability. 100 hits with probability 1.0 mean nothing. 10 hits with probability 0.25 mean a lot.

Roger