Sagarin ODU by 5.7
Massey ODU (73%) 35 - 28 MU (27%)
0/3 Sims ODU wins 7,194 of the sims at an average score of 35-26. ODU raw probability is 82%
ESPN FPI ODU 62.9%
It is extraordinarily rare for my sheet to have a widest margin, but this is an unusual year.
Usually, one blowout win or loss isn't enough in the long run to skew a team's standings in a prediction formula. This season has forced me to look at my own system and devise a more accurate way of producing team inputs for the big formula. I've devised a new sub-routine for team stats that puts more weight on more recent outcomes, a sort of "what have you done for me lately?" approach that rewards good scoring margins closer to the upcoming game and leaves early-season money games behind. The result above is using the new formula and the result below is using the old formula:
While it doesn't look like much, the fact that MU's offensive production over the last three games has gone from 27 to 21 to 14 while ODU's has remained higher (36, 24, and 31). On the flip side, MU's defense has been relatively consistent (21, 27, 24) while ODU has been much less so (16, 59, 21).
Nevertheless, I think it paints a more accurate picture of the Herd at this point. Comparing the two teams' raw numbers:
Avg winning margin using raw mean system:MU 30-35 (-5)
ODU 32-29 (+3)
Avg winning margin using weighted mean system that blunts the effect of early blowouts:MU 24-33 (-9)
ODU 32-30 (+2)
So, Marshall is rewarded for their improved defensive play, but penalized for their decreased offensive production. ODU has been inconsistent, but consistently so - therefore the system does not really reward or penalize their lack of a trend.
Sad to say, this widened the margin for the game prediction by a couple of points. I have no idea what else will need to be tweaked to compensate for they way the scores are being handled, but this is probably a good season to get experimental.