Game Probabilities Week 7

Game probabilities for week 7 are up at the New York Times. This week I take a look at the Patriots Jets matchup and some surprising numbers.

A purely statistical approach is (deliberately) ignorant of things like the identity of the quarterbacks. It doesn’t know about Super Bowl rings or supermodel wives. It doesn’t know who is a rookie and who is a certainty for the Hall of Fame. It doesn’t know that Patriots linebacker Jerod Mayo is out for the regular season with a chest injury.

  • Spread The Love
  • Digg This Post
  • Tweet This Post
  • Stumble This Post
  • Submit This Post To Delicious
  • Submit This Post To Reddit
  • Submit This Post To Mixx

17 Responses to “Game Probabilities Week 7”

  1. Anonymous says:

    I see that you have the Cinci/Detroit game as .50/.50. Is this really dead even or do you have one team a slight favorite?

    Thanks for all you do!

  2. Anonymous says:

    Oh man you are going to get blasted for picking Indy as a favorite (albeit a slight one) over Denver.

  3. Anonymous says:

    Have you ever run this against the winning ML to see how you fare?

    Also, HOU last week? 83% losing by a large margin....although I believe they outgained STL.

  4. Doctorjorts says:

    Call me a homer, but I'm shocked (SHOCKED) that the Cowboys have a lower win probability against the Eagles this week than they did against the Broncos two weeks ago. What gives??

  5. Anonymous says:

    I am not affiliated with this blog, just mentioning as it seems relevant:

  6. Dan says:

    Did the formula change? For instance, Miami has a GWP of .42, and Buffalo has a GWP of .39. With home field advantage, the old formula has Miami winning 60.7% of the time. However, you published Miami winning 53%. Did I miss something?

  7. TPM says:

    @Doctorjorts - remember that Brian's model is taking into account how the teams did last week. Although Dallas beat Washington by 15 points, that was mostly due to special teams (Harris' punt return TD and 90-yard punt return which set up a Dallas TD) and turnovers. The Dallas offense, and the passing game in particular, wasn't really that great. On the other hand, Philly had a SOLID offensive outing against Tampa Bay (who, admittedly, aren't that good), particularly their passing game.
    The model most likely gives more weight to the offensive performance, which is more consistent from week to week, than the special teams and turnovers.
    Whether or not the model is correct in analyzing things this way is another matter, but that's why I think you're seeing what you're seeing.

  8. Mike says:

    Doctorjorts - The biggest driver is probably home field advantage. Dallas was at home versus the Broncos and away versus the Eagles. That's a good 15%-20% swing in win probability right there.

  9. TPM says:

    @Mike - I was also going to say that...really.

  10. Marv says:

    The 49er game makes me nervous. Kaepernick's SR (43%) is extremely low... I keep waiting for the game where Kaep doesn't hit so many bombs and the niners fail to score. Tennessee has some good young DBs and they're playing at home, it could happen this week...

  11. Anonymous says:

    Marv - yeah man, big 49er fan here. Tennnessee has been a tough team this year and SF's addiction to the running game might bite them in the arse. Its fortunate that Locker is out otherwise it might be a tight game!


  12. Unknown says:

    Hey, Brian. How does your model incorporate (or not) home field advantage?

    More importantly, I appreciate the pure statistical analysis, but I want a coach (and an analyst for that matter) who understands the limitations of the pure statistical analysis. You correctly point out the influence of randomness, particularly in a low-data environment, and I also think there are important externalities to consider. My biggest beef in this area is the 4th down debate. I think it is much more context-dependent than the pure statistical analysis allows. For example, I think it was 100% correct for Belichick to go for it in the 2009 game against the Colts because he was playing against Peyton Manning (although the potential psychological impact of "not trusting" your defense is a counterveiling externality). But if I have a late, small lead against, say, Blaine Gabbert, I would always punt on fourth down in my own territory. As you say, the pure stats do not factor in context and so are of limited utility, and I want someone who understands that and uses the stats as the foundation to make context-based decisions.

  13. mitch says:

    According to the model the Colts -1.5 over Bronco's.

    The model is suggesting to fade the Bronco's again this week It's won the past 2 weeks fading Denver as I posted those plays.

    Model says, 49ers -3.5
    Eagles -3.5
    Miami -4
    KC -8

    The model is suggesting a fade on Miami, however I would disagree with that one because of regression, which is coming in on Miami and is not accounted for in the model to the degree it's coming.

  14. mitch says:

    Since we're talking Pats/Jets the model says Jets -5

    Suggesting the Jets + the points.

  15. Unknown says:

    @mitch - on taking that Jets one as well. I could easily see Brady and Co. Beating them by 2 TD's, just because it's Brady and Co., but my model says the same thing, so I'm sticking with the numbers.

  16. Anonymous says:

    Let me echo @Dan's question. How does one produce these win probabilities numbers starting from GWP (or should I be starting from something else)?

    I'm unable to reproduce them using the formula from

  17. Brian Burke says:

    To answer the questions above, please read the article titled How the Model Works.

Leave a Reply

Note: Only a member of this blog may post a comment.