Don't Believe Your Eyes

Both ESPN and Football Outsiders have released their Week 2 power rankings. In the last seven days, the average team in ESPN's poll has seen their ranking change by 4.13 spots; in FO's poll, the average change has been 2.06.

This quantitatively tells us something I think we already intuitively knew: that people generally overreact to Week 1 results. The Falcons--up from 32nd to 25th in the ESPN poll--managed to beat the Lions at home, so clearly they're a totally different team than the one that was outscored by 155 points in 2007. The Vikings lost to the Packers, in Lambeau, by five points, and they drop from 11 to 16. I guess the folks at ESPN thought Tavaris Jackson was going to have an MVP-caliber year before Monday night's performance?

These are, of course, complete overreactions. Week 1 certainly isn't meaningless, but it's only one additional data point, nothing more. FO weights the Week 1 DVOA as just 10%, with the preseason projection accounting for 90%. So the Vikings, who were #5 in the preseason, stay there, and the Falcons move up only two spots, from 32 to 30.

Not surprisingly, the betting public is much closer in ideology to ESPN than FO. The Week 1 winners, despite being favored by an average of less than a quarter of a point, are currently averaging 56.4% of the action on Wagerline. The most extreme example of this are the previously mentioned Falcons. Playing in Tampa and getting nine points, they're getting 64.9% of the action, a huge number for such a big underdog.

Another interesting aspect of this is how it has played out in recent years. Over the last two season, teams that have won straight up in Week 1 are 19-11-2 against the spread in Week 2, an impressive .633 W%. That can easily be chalked up to random variance; anything can happen in a sample of just 32 games. Looking at the same stat from 2002-2005 confirms this; Week 1 winners are only 30-33-1 (.476) ATS in Week 2.

It makes you wonder if the public is overreacting to their recent success of betting on Week 1 winners. In '06 and '07, the average Week 1 winner was favored by 2.8 points in Week 2. Since the Wagerline public tends to bet on favorites at about a 60/40 split, and the teams were favored by more in those two years than in '08, we would expect a higher % of bets on those 1-0 teams. In fact, they only received 55.8% of the action, slightly lower than this year.

That may be a random fluctuation, but the larger point is clear: both the media and betting public vastly overreact to Week 1. You should probably take it with a large grain of salt; after all, it is only one game.