New Computer Ranking Update
I just wanted to let everyone know I’m making great progress on my own computer rankings. My first run at it was with Boise State. I took their season average in points scored and allowed minus the game against TCU, then took TCU’s season average in points scored and allowed. I then took Boise State’s season average and compared it to each of their opponent’s season average, minus their game against each other. I determined that Boise State should expect to score about 1.5 times their opponent’s season average in points allowed. They should also expect to hold their opponent to about 65% of their season average in points scored. Interesting results.
As it turns out, by using those numbers, the expected outcome of the game would have been a Boise State loss of 19-26. But they won 17-10. That counts as a quality win, seeing as how they did significantly better than expected. Now I did the same for each game, expected results next to actual results. Another quality win was Hawaii. Based on the numbers, Boise State was expected to win that 42-12. They won 54-9. Now what I have to do is run a chi-square test to see if the difference in actual outcome and expected outcome is statistically significant.
Assuming it is, then it counts as a quality win for Boise State. Then when you want to compare teams, you plug in each team into the matrix that gets created. Let’s say you plug in Alabama against Hawaii and the expected outcome is 48-14. Well, by the numbers, Boise State played better than you would have expected Alabama to do. So yes, Alabama beat better teams during the regular season. But are they better? Yes, Texas beat Nebraska, but wouldn’t you have expected a team ranked #2 in the country to have done better?
Using this data will help me determine not just who beat better teams, but who actually played better. You expect good teams to play each other to a close game. You expect good teams playing bad teams to blow them out. The question is by how much and whether or not the good team met or exceeded what they were expected to do. Teams who exceed expectations each week should be ranked higher. Teams who just win but fall short of expectations should be a little lower. Teams who fall way short of expectations and lose should be even lower. I’ll keep you all updated on my progress as I continue with this project.
Comments
No comments yet.
Leave a comment