## NFL Week 3 Power Rankings

Everyone loves power rankings!  Otherwise, how do you know who the best team is?  So, here is my version of them (here are ESPN’s if you’d like a comparison).  First, when I say week 3, I mean going into week 3.  That means there are two games worth of data going into these rankings.  That isn’t a lot, so I expect the numbers to move around and the order to change in future weeks, but it should still be informative and fun.  Second, this works off of the Luigi win probability model.  I chose the win probability model over the point difference model because people are typically more interested in who will win, not how much they’ll win by.  Third, since the model is set up to compare home to away teams, for computing the power rankings I treat every team as if they were playing at home.  Thus, to give a sense of how good teams are, I also compute the power ranking of an average home team from the past six seasons and the season so far (e.g., every game in the data set).

And fourth, I have a ranking for each team’s “offense”, “defense”, and total.  I put offense and defense in quotes because offense is really things that the team did, like acquire passing yards and rushing yards but also interceptions thrown and times sacked, while defense is things that the other team did.  It’s a little fuzzy as to if team A threw an interception or if team B intercepted the pass, if you see what I’m saying, so to a certain extent ‘offense’ and ‘defense’ mix both offensive and defensive plays; the main idea is that offense refers to stats acquired by that team’s offense, including bad things, and defense refers to stats acquired by that team’s opponents.  Anywho, here are the rankings!

Things to note in the table (once you’ve clicked on it to make it more readable): offensive values should be positive, and larger is better.  Bigger values in the equation mean a better chance of the home team winning.  Similarly, defensive values should be negative, because the more stats the opposing team acquires the lower your chances of winning, but a better team will have a smaller (less negative) values (they will limit the yards gained, and the efficiency of the yards gained, of the opponent and force them into more turnovers).  The total is simply the addition of the offense and defense scores, and again bigger is better.  The average home team wins 57% of the time against an average away team according to the model, which is roughly the correct proportion.  So teams above ‘AVERAGE’ would beat an average away team more than 57% of the time, and those below would win less often.  The table is sorted by total score.

A couple of random observations.  Houston is far and away the best offensive team; with even an average defense they would beat an average away team 78.5% of the time.  Seattle seems to be an odd choice for the number three team given how they were just manhandled by Denver, but they’re being driven by their great game in week 1.  The 49ers did nothing in that game, and the Broncos actually ran very poorly in the week 2 game, so Seattle’s defense looks pretty good right now; the offense also moved well in both games, they just had 4 turnovers against Denver.  Chicago, despite not running the ball well at all, has great passing efficiency and run defense and has forced three turnovers per game so far.

At the other end of the spectrum, Buffalo is the worst offense in the league, which is unsurprising if you know anything about the Bills.  With an average defense (which the Bills have, roughly), their offense would give them only a 26% chance against an average away team.  Pittsburgh and the Jets, unsurprisingly, have the best defenses while the Eagles and the Colts are the worst.  Despite the final score, the Giants moved the ball ok against the Colts and the Texans obviously had their way with them; the Eagles gave up all sorts of yards to the Packers and the Lions.  So these all seem reasonable.  The Saints are a surprise, showing up below average.  However, if you watched both games (and you might have since they were both on national TV), you saw that the Saints have struggled so far.

So there you have it, your week 3 power rankings.  The current best team in the league, the Bears, would have an 82.2% chance against an average away team; the worst team, the Bills, would have an 27.8% chance, so in one game it’s still true that anything could happen.  Looking ahead to this weekend, the Patriots should have their way with the Bills while Carolina and Cincinnati (and Oakland-Arizona) hold an exercise in futility.  The Saints might be surprised by Atlanta, and there should be some good games in Chicago-Green Bay, maybe Seattle-San Diego (we’ll at least see how good Seattle really is), and Tennessee-Giants.