Week 9 Power Ratings and SEC slate

UPDATE to the model:

Added turnover rate at the suggestion of Quicksilver. PFF found that 50% of turnover-worthy throws by QBs are intercepted. And 1/3 of all interceptions come from passes that are not classified as turnover-worthy. This is basically saying that more than half of interceptions are caused by bad throws by the QB and from the perspective of the defense are considered "luck".

Football Outsiders has determined that forcing a fumble is skill-based, but actually recovering the fumble is around 50%.

Because of all this, Bill Connelly also uses 50% in SP+ ratings.

So what this means is teams that have a high net turnover margin will be docked - because they are considered to score points off lucky turnovers. But it's not a blanket increase or decrease, it's still by game. In football, turnovers correlate to 5.7 points per turnover. A turnover points value of 2 per team (because 50/50) ended up being the most accurate for predictions. So, here's an example of how it's actually applied in my model now.

Kentucky - 0 turnovers - 15 offensive points, 6 defensive points
Oklahoma - 4 turnovers - 13 offensive points, 0 defensive points

the model says Kentucky earned two of those turnovers and the other two it got lucky, so it reduces Kentucky's offensive score by 4 (2 x 2), and increases Oklahoma's offensive score by 4 (2 x 2). So the model basically sees the game this way:

Kentucky - 17 points
Oklahoma - 17 points

And that's how credit for the game is then dispersed. So it hurt Kentucky and helped Oklahoma.

Teams that have a high negative turnover margin, it will help (Oklahoma, Oregon). Teams that have a turnover margin around zero, it will be insignificant (LSU, Ole Miss, BYU). Teams that have a high positive turnover margin, it will hurt (Kentucky, ND, Miami, Missouri, Nebraska, UCLA).

If you see a team that is rated higher than expected (Oregon) and they turn the ball over a lot - this means they are keeping games close even though they give the ball away a lot. That's why it helps their power rating.

If you see a team that is rated lower than expected (Kentucky) and they have a lot of takeaways - this means they are letting other teams keep the game close even though they have a lot of takeways. That's why it hurts their power rating.

Adding this IMPROVED backtesting MAE by 0.1 points - which is a considerable gain.

LSU jumped to the top because of yet another dominant performance over Auburn and because ND and Miami both fell due to the turnover luck.

If you want to cry about LSU, they have high expectations right now. They are currently favored over Bama by 8(h), Texas by 9.5(h), and Tenn by 7(a). If LSU doesn't live up to that, they will quickly fall. This will all work itself out in the end.

Submit your ignorant tears below.