Filed under:

# Do bye weeks actually help?

There are two steps involved in the release of a given year's SEC football schedule. Step 1: It is released. Step 2: Alabama fans note how many of their upcoming opponents have bye weeks before facing Alabama.

To be sure, it seems there are always examples. This past season, three consecutive Bama opponents -- Texas A&M on Oct. 17, Tennessee on Oct. 24, and LSU on Nov. 7 -- played the Crimson Tide after a week off. (Bama also had a bye before LSU.) The opponent after that, Mississippi State, even had a couple of extra days off because of a Thursday night game the week before.

But does this actually matter? Looking at something like bye week effects is so based in anecdotes that you can see whatever you want to see. I've always wondered if the rejuvenating value of a bye week is offset at least a little bit by the fact that you fall out of your week to week rhythm. What do the numbers actually say about the effects of bye weeks?

To search for an answer here, we're going to look at percentile performances and a team's performance vs. the spread. You can find out more about the percentiles hereI use them in my in-season team profiles and offseason preview series. It boils your single-game performance down into a single, opponent-adjusted figure. In theory, using percentiles to compare teams' performances after a given number of days off could work out pretty well.

The verdict? An extra week off might have been worth a couple of points in 2015.

 Quality of Performance by Days Off* (2015) Days Between Games N PercentilePerformance Week to week change inpercentile performance vs. Vegas 5 days 30 47.5% +1.2% -2.1 6 to 8 days 1183 52.6% -0.5% +0.1 9 to 13 days 117 47.1% -0.4% -1.9 14 days 81 54.9% +3.3% +2.3 15+ days 14 45.8% -7.1% +0.7 9 or fewer days 1281 52.2% -0.2% -0.1 over 9 days 144 50.9% -1.3% +0.7 * This sample filters out all Week 1 performances and bowl games. For performance against the Vegas spread, games against FCS competition are also filtered out.

Some thoughts:

• I provided a few different ideas for data points here, and sometimes they conflict. The five-day breaks (a.k.a. the Thursday night games), for instance: teams play at a pretty low 47.5% level, but they improve by 1% over their previous performance? While playing two points below what the spread projected? The sample size for those games was small enough that a few odd performances could sway the numbers.
• The sample was definitely not too small for the 6-8 day window, and as we see, those numbers are pretty stable.
• 9-13 days would basically be either playing on a Wednesday, Thursday, or Friday after getting a bye week or playing on Saturday after a Tuesday, Wednesday, or Thursday game. There are a lot of MACtion games in this sample, in other words, so the fact that the average percentile performance is below 50% might make sense. It's interesting, though, that teams pretty drastically underachieved compared to the spread. Generally speaking, those midweek games don't do favors to the overall quality of the product.
• The most important piece of this is probably the 14-day sample. That's your standard bye week. And teams played at a high level, improved by more than three percentile points, and overachieved by more than two points compared to the spread. The sample isn't huge here, but that's a pretty definitive bump. It's not worth a touchdown or anything, but in the grand scheme of things, two extra points is quite a bit.

This is only a one-year glimpse, and a lot of the numbers are pretty inconclusive. But at first glance, it does look like the bye week could be worth something to you ... even if simply means losing by a little bit less to Alabama.