Andy Reid After Bye Week

Disturbed by the news that Andy Reid is 11-0 all time coming off a bye week, reader Dan Svirsky set out to investigate if that was lucky or a trend.

This week the Colts face an Eagles team fresh off a bye. As some have noted, Andy Reid’s teams have gone 11-0 after bye weeks since 1999. So, are the Colts in trouble? This analysis looks at two questions:

(1) Are teams really better after bye weeks?

(2) Is Andy Reid specifically better after bye weeks?

To do the analysis, I got data from 3,041 games over the past 11 years. For each game, I looked at three variables: Was the team playing at home? Was the team coming off a bye? What was the DVOA spread of the game (or the difference between the team’s weighted DVOA and the opponent’s weighted DVOA)? Then, I ran a test to see what happens to a team’s margin of victory as these variables change (I used a multiple regression test, for the stats nerds out there). So, on to the results…

(1) Are teams better after bye weeks?

Based on this analysis, the answer is a pretty clear yes, which I didn’t expect, to be honest. I assumed this was just conventional wisdom that wouldn’t bear out. But this is what the numbers said:

Teams coming off a bye week had a margin of victory (or defeat) about 3.43 points better.

Teams playing at home had a margin of victory (or defeat) about 2.96 points better.

Teams with a weighted DVOA 10% better than the opponent’s had a margin of victory (or defeat) about 3.07 points better. So, if Team A with DVOA of 10% plays an average team, then, holding other factors constant, Team A should win by 3 points on average. So what are the takeaways?

• The bye week does matter

Teams coming off a bye do better—whether because players are rested or because coaches have more time to prepare. Which factor is more important? To get an idea, we can look at divisional playoff matches, where one team has a bye, but neither team has two weeks to prepare (though you could argue that a team that has a first round bye could start to prepare for one of its possible opponents…I personally don’t see this having a huge effect, but maybe Bill Belichick or Tony Dungy can weigh in if they’re reading this). Looking at just divisional playoff games, the home team does about 3.8 points better. If you assume that the “home team effect” is the same in the regular season and postseason (maybe a stretch, but you could argue it both ways), then it seems that the value of resting players for an extra week accounts for about 1 point. Since the bye week advantage is 3.43, we could guesstimate that the bye week advantage is 1 point rest, 2.43 points preparation. This is really, really fuzzy math, though.

• Vegas knows what it’s doing.

Home teams usually get a 3 point bump in the spread from what I understand, and that matches the data almost exactly. Pretty impressive.

• More on home field advantage

Here’s a table showing a few teams and how many more estimated points they get when playing at home:

Because the samples for each team is smaller (only 11 games), there’s less confidence for these numbers. Either way, looks like the Colts do fine whether at home or on the road. As you could guess, the Seahawks have a huge bump being at home, and the Broncos and Eagles do a bit better.

(2) Is Andy Reid better after bye weeks?

Again, I was surprised to find that Andy Reid teams do, in fact, perform better after bye weeks—even better than other teams after bye weeks. The table below shows each post-bye week game that Andy Reid has played, along with the actual margin of victory and the expected margin of victory from the model. (Here’s a quick example: take the 2009 Tampa Bay game. The Eagles were at home–good for around 3 points in our model. They were coming off a bye—good for 3.43 points in our model. And they had a weighted DVOA 52% better than the Bucs—good for roughly 15 points. Add it all up, and our model would estimate that the Eagles win that game by about 19.5 points):

In sum, Andy Reid does about three points better than you’d expect when he’s coming off a bye week, though that effect hasn’t held true for the past three years. He actually beat two Super Bowl champions in his first four years (the ’02 Bucs and the ’99 Rams). Because the sample size is small, we don’t have a ton of confidence for this result. You could argue that Reid is only average after a bye week but he just got lucky, or had some fluky results. This could be the case, but statistically, even with this sample size, you’d get this three point bump just by chance only 1 in 10 times.