2014 Super Bowl Prediction

In less than 24 hours, we’ll have a Super Bowl between two of the best and most evenly-matched teams in NFL history.  I think the general consensus is that the main determination will be if the Patriots can stop Marshawn Lynch and if the Seattle defense can hold up against the Pats; whichever side does the better job on those two issues will win the game.  But there are always numbers to look at!  Let’s break down the game.

A Quick, Bayesian Note on Deflate-Gate

Neil Paine wrote an article rounding up a bunch of the posts on the issue of if the Patriots are or have ever been Communists let the air out of their footballs.  I’m glad he did, because I was thinking about it but was too lazy to do it.  There is a lot of good information in there, especially if you go through all the links, but I wanted to focus on the “the Patriots stopped fumbling” portion of the discussion.  In my last post, I briefly mentioned that you could take a Bayesian angle to the question of if the Pats cheated in the NFC title game: do you think that they would cheat in such an obvious manner?  You can extend the same thinking to the issue of if the Patriots have been cheating since 2007 to account for their low fumble rate:  what is the probability that the Patriots have been treating their footballs since 2007 and have never been called out on it? 

NFL Conference Championship Results

With there being basically a dead weekend before the Super Bowl (does anyone really watch or care about the Pro Bowl?), I put the conference title games on the backburner until now.  As predicted by essentially everyone, the two favorites won (that’s why they’re favorites).  As not really predicted by everyone, one won by the most fortunate of circumstances while the other is now deflecting criticism that it may have cheated.  Let’s flash back all the way to… one week ago.

NFL Conference Championship Predictions

Tomorrow we get two games featuring probably the three consensus best teams in the league this season, and the Colts.  If we were back in September, you’d probably have the Broncos instead, but Indy handled that directly.  You might have the 49ers or Saints, but they were fairly obviously out of the ‘top team’ race during the season.  If you go back to 538’s Elo ratings at the beginning of the year, that’s basically what you have; the exception is the Packers, who were artificially low at 17th due to Rodgers’ injuries last year (they still made the playoffs).  The current ratings have the Colts at #4, so according to Elo we indeed are seeing the four best teams, but that’s because they won last week.  Before the game they were 7th, and had Baltimore won I think the Ravens might be ahead.  But at any rate, we have a good group of teams to watch.

NFL Divisional Round Results

Well, we avoided having an exact repeat of last year’s conference title games.  Instead we only have two repeat teams.  Seattle has made it again but will face the Packers instead of the 49ers, and the Patriots are back but playing Peyton Manning’s old team instead of Peyton Manning’s current team.  But before I get ahead of myself, let’s see how the models did this past weekend.

NFL Divisonal Round Predictions

Since the Lions were knocked out while simultaneously demonstrating that the officiating system is broken while the NFL does nothing to fix it, it’s unclear why there are still playoff games.  After all, I just wanted to watch the Lions.  And it seems perfectly reasonable to expect that the referees will get things wrong and affect the game.  Maybe they’ll change the rules next year…. again.  But I guess some fans would be disappointed if we didn’t play out the string, so we have another pair of games each day this weekend.  This round we have division champs coming off the bye, so not only do they tend to be pretty good, but they’re also rested and at home.  Let’s see how things might play out.

NFL Wild Card Results

The Lions game just ended, so it’s still a little tender.  But here’s how the model predictions did for the wild card round.

Using Numbers to Say What You Want

I’ve been more or less enjoying the articles over at 538, and not just the sports ones.  I also  started a new job as a researcher at the FAA back in the fall, so it jumped out at me when I saw a post titled “A Deadly 2014 For Air Travel Has Reversed Recent Safety Gains“.  My job isn’t to run numbers on these kinds of things, but I get emails with FAA news, and the ones I saw seemed positive.  Crashes were down.  Why is 538 telling me that safety gains have been reversed?  It’s because it picked one number to base the title on instead of all the other ones it mentions in the article.

NFL Week 17 Results and Season Prediction Summary

The regular season is in the bag, and before moving on to the playoff predictions (and how sad they look for the Lions), I need to wrap up the models’ performance this year.  It will be a sad tale of woe, but there is no avoiding it.  As I noted when making the predictions, week 17 is a tough one because the model doesn’t know who is in the playoffs and who isn’t, and thus who is resting and who is playing hard.  Let’s see if it led to a bad week.

NFL Week 17 Predictions

Week 17 of the NFL season is a tough one for models and predictions in general.  Some teams have their playoff spot completely determined, like the Pats, and so have nothing to play for and have to balance keeping sharp with avoiding injury.  Some teams are completely out of the playoffs and have nothing to play for except a slightly different draft spot, like Buffalo.  And some teams, obviously, have playoff spots and/or seeds to play for.  My models don’t take any of that kind of thing into account, and even if you try to it can be hard to know in advance how long starters will actually play, etc.  So for this week’s predictions, I’m going to go through each game and make some guesses on what the models say versus the lines.