“I guess it’s a residue of the circular reasoning with SRS, but if you sum up the DIFF for each team, they all net out to within 0.2 of zero.”

Something like this – I don’t understand the maths behind it but that’s the way it works. The interesting one is if you take the standard deviation of this (the DIFF) you’ll usually find after about 6 or 7 weeks of the NFL season it’s somewhere around 12 or 13. This backs up the paper by Hal Stern from 1991 “on the probability of winning a football game”. Neil had an epic post/comment on this a while ago where he calculated the standard deviation over about 20 years worth of games and proved on average it is slightly lower than the recognised value from the Stern paper of about 13.86 (which makes a small percentage difference in outcomes of predictions).

Interesting one and a good post, as I think it comes down to that whole backsight and foresight projection. I just adjusted my spreadsheet back to week 5 (the furthest back SRS figures on here are through the Sunday games of week 6) to remember how well Seattle were ranked at the time and I think it comes down to the fact that we weren’t believing what the figures told us essentially (in the same way that I shouldn’t have believed what the figures told me about Arizona in week 5). I think we all felt Seattle were going to get found out. The SRS figures on here through the week 6 games actually only had a two point difference between the Pats and the Seahawks and 9 weeks on from that, those figure looks valid (difference is only 0.4 different).

The question is I think is; is it more valid to look back on the figures now as a backsight projection of your current figure from week 15 onto the team at the time (when arguably we all have a lot more faith in all of the numbers for all of the teams through 15 weeks) or to take the figure for the team from before the game (when arguably there was more doubt about some of the figures – e.g. Arizona or Miami)? The best example I think is the Miami, Tennessee game which still provides the biggest surprise but if you take the figures from before the game (which again I shouldn’t have believed….) the surprise was even bigger (when the figures had Miami as a substantially better team).