paper: OPR Match Predicter

[kendalls]

Thread created automatically to discuss a document in CD-Media.

OPR Match Predicter (http://www.chiefdelphi.com/media/papers/2483?) by kendall

Please forgive the spelling error; Chrome's spell checker lied to me.

[kendalls]

So I've been comparing the predicted results with the actual results of today's matches in the Galileo division. I've determined that the most accurate OPR for a team is the one from the tournament where the team got the highest OPR. For these OPRs, I got a 27% error in predicting match scores and a 22% error in predicting the winner of a match. For normal (average) OPRs, I got errors of 35% and 25% and for 1114's normalized OPRs I got errors of 39% and 30%.