Game probabilities for week 13 are up at the New York Times' Fifth Down.
This week make a cursory comparison between the 49ers' Colin Kaepernick and Alex Smith.
- Home
- predictions
- Weekly Game Probabilities
Weekly Game Probabilities
By
Brian Burke
published on 11/29/2012
in
predictions
Subscribe to:
Post Comments (Atom)
Your table at the NYT has a 51% chance of the Pats winning this weekend. Is that right? Is this a home-field thing, injuries to the Pats, historical experience?
It's their pass defense.
I agree. Very weird, but teams just aren't taking advantage of the NE defense enough.
NE has a "net" of 141.4 EPA between their offense and defense on the season, while Miami's is -39.9 (offense minus defense for both sides, not sure if your stats work this way). This is a difference of 16.5 EPA per game. PFR's SRS has the teams 17.2 points apart. The pass defense, which is bad, has been a part of this team all season, and they have STILL built up such a massive difference.
It isn't like NE has been starting four hall of famers in the secondary and they all retired last week.
Again, I'm not sure if you can just net out the EPA differences, but this game sure FEELS like NE should be massive favorites. Even as bad as the NE pass defense is, MIA isn't that much better. The gap between NE's pass offense, and Miami's pass defense is 185.4 points (142.5 + 42.9). Going the other way, the gap is 70.1 (3.0 + 67.1).
Please don't take this as a knock on your stats, such as EPA. I think this is all brilliant stuff. It just seems a little incongruous with the win probability.
New England has won 21(!) more turnovers this season than Miami, and I think the ANS model tends to expect a reversion to the mean there.
Nate: just an interesting fact with no necessary predictive ability but since 2001 (Brady+Belicheck) the Pats are +132 in turnvoers for an average of +11 per year with only one negative turnover differential season -6. Most of the differential has come from their ability to avoid interceptions (Brady) + high emphasis on short passing game as well as the ability to take them away on defense.
I've been saying this a few times but would be really interesting to have a case study post on NE. It really seems they're one team that has consistently been breaking the model. Could be just a statistical quirk or could be that they're actually playing a different game.