Now that the season is over, there are only a few things to do until the playoffs start. I’ll go with the final regular season power rankings and the season predictions, and take a look back at some predictions made along the way.
Here are the final power rankings. They obviously feed into the playoff predictions, so you can look for your favorite teams/Tebow.
The Saints finish the season as the top team in the league, by way of having an average defense while the Packers’ is below-average. The Eagles managed to finish at number 4 despite going 8-8. That may seem odd, but also note that they had a positive score differential, actually 5th best in the NFC (they also would be 5th in the AFC). That means that in general they lost close and won big, which is what good teams do. Denver is notable for being a playoff team that the model thinks is fairly terrible; they’re 26th and the only below-average playoff team this year. As a Lions fan, I’ll point out that they really did themselves a disservice by losing to the Packers. Instead of a first-round game against the Giants, which is pretty winnable, they have to go to New Orleans and will likely lose again.
And here are the season predictions.
I’ve noted before that the best teams tend to be the luckiest and the worst teams tend to be the unluckiest. Luck for me is taken as the difference between the predicted win and actual win columns. The Texans show up as unlucky here; that is likely at least in part because of all the injuries they suffered. Their team stats expect them to be better than they were at the end of the year. At the other end we have the Rams, Colts, Vikings, Bucs, and Browns, all teams that were bad but could have been better.
I wanted to look back to see how teams have changed over the season. I did this before by plotting each team’s rating over the course of the season, but it makes for a messy graph. Instead I decided to look back at the ratings from week 7; that’s about when teams start to really look like their year-end performance. Back then, the top 5 teams were the Packers, Patriots, Saints, Ravens, and Cowboys; the worst 5 were the Rams, Colts, Jags, Browns, and Chiefs. Dallas dropped off, and the Bucs really dropped off and got into the bottom five at the end of the season, but otherwise those are spot-on. The correlation in ranking between week 7 and 17 is .87, which fits with the general impression. The absolute strengths may change a bit as well, which is important, but we roughly know which teams are good and which are bad about halfway through the season. Doing the same thing with expected wins, the correlation between expected wins at weeks 7 and 17 is .977, and the correlation between week 7 expected wins and final actual wins is .737. That isn’t too different from the correlation between actual and week 17 expected wins, which is .822.
It turns out that I made my first playoff predictions in that week 7 post, which is quite the coincidence since I had forgotten my post about when teams become ‘themselves’. I thought the AFC playoff teams would be the Pats, Texans, Chargers, and Bills but noted that the AFC North could send three playoff teams, which is what ended up happening. In the NFC I tabbed the Packers, Saints, and 49ers along with the Lions, Bears, NFC South, and NFC East (things were tight at that point). Those three division leaders ended up winning, obviously, and the Lions made it along with the Giants and Falcons. The Bears fell off due to injury, the Falcons were better than expected, the Panthers stayed about on target (which was too low), and the Bucs just fell off a cliff.
So all in all I think the predictions went pretty well. Apparently by week 7 you can get a fairly solid lock on what the league looks like. I overestimated the Bills and underestimated the Falcons, but that was true all season long. Maybe next year I’ll make playoff predictions in week 5 to add an element of uncertainty.