Personally, I prefer this question to be addressed by the predictor - he should
set limits on his/her prediction which he feels are justifiable. If the limits are
not correct the prediction is considered a miss. I don't feel our job is to make
the predictor "feel good", but to evaluate a prediction as it was made.
However, a case can be made for near misses. I will have to give that some
thought and get back to you.