Prime Time Performance in the NFL

Posted on January 23, 2007 by

3


The myth of the prime-time performer is one of the stories we tell in The Wages of Wins. Specifically, we find that in the NBA, players do not systematically play better in the playoffs. This finding is based on a study of regular season and post-season per-minute Win Score. Our examination demonstrated that players did not consistently post better numbers on NBA’s biggest stage. In fact, many stars like Michael Jordan tended to play a bit worse.

Here is what we say about Jordan in the book (p. 157):

Of course one might expect performance to decline in the playoffs. Teams play better defense and at a slower pace in the post-season. So naturally players accumulate fewer points, rebounds, etc. To see how big a decline we can expect, we collected data for every player in every post-season from 1995 to 2005. On average, per-minute Win Score declines by .03. Even when we take this average decline into account, though, Jordan still doesn’t consistently get better in the playoffs. Specifically, if we add .03 to Jordan’s playoff performance, adjusting for the expected decline from the slower pace and better defense teams play in the post-season, Jordan’s production in the post-season is still worse than his regular season output in eight of the thirteen seasons we examined.

As we note, this finding is not limited to Jordan. We find that many “star” players tend to play worse in the playoffs. It’s important to note, as we emphasize in the book, the decline we observe is fairly small. And it’s still the case that Jordan was still often the most productive player on the court in the post-season. He just wasn’t quite as good as he was in the regular season, where he got to play teams that were lottery bound.

Prime-Time in the NFL

We did not look at the NFL playoffs to see if a similar pattern would be observed on the gridiron. This year, though, we now have data on 10 playoff games. Although 10 games is hardly an adequate sample, let’s pretend it’s enough and investigate whether any quarterback has raised his game in the NFL’s post-season.

Table One: NFL Quarterbacks in the 2006-07 Post-Season

The above table reports QB Score per play in the regular season and playoffs for the 12 quarterbacks who led a team in the post-season this year. For nine of these players, per play performance has declined in the playoffs. So we have a bit of evidence that NFL quarterbacks, like NBA players, play worse in the playoffs. Before we get at the problems with this analysis, let’s look at the two Super Bowl quarterbacks – Rex Grossman and Peyton Manning.

The much maligned Grossman appears to have raised his play in the post-season. Of course, this might be because Chicago appears to have adopted a strategy of trying to limit the likelihood that Grossman would commit a turnover. In the regular season Grossman averaged 4.76 turnovers per 100 plays. In the two playoff games he has cut this rate to 2.9 turnovers per 100 plays.

While Grossman has – at least in terms of the non-conclusive numbers we see – played a bit better, his opposite in the Super Bowl appears to be playing worse. Of the six quarterbacks who have played more than one game, Peyton Manning has suffered the biggest decline. In the regular season he was the most productive quarterback in the NFL. In the post-season, only Steve McNair and Trent Green have offered less per play. Of course, these two quarterbacks faced the suddenly stellar Colts defense. In sum, although Manning has not played well in the playoffs, the quarterbacks who have opposed him have always managed to play a bit worse.

Problems and Problems

What are the problems with this analysis? Well, I can see two issues immediately (and perhaps you can see others).

  • The sample size is much too small. In our study of the NBA we required each player to have played at least 10 games in the playoffs in a single season to be included in our analysis. Our entire sample of NFL quarterbacks is 10 games. And that’s far too small to tell us anything.
  • The sample is also biased. Yes, the majority of quarterbacks played worse in the playoffs. But of the nine who played worse, six were on teams that only played one game. If we restrict our sample to those who at least played for a winner in the playoffs, we see that three signal callers played better and three played worse.

Once we consider the issue of sample size and sample selection, we see that we do not have much of a story.

So what have we learned? The analysis we offered of the NBA is (I think) fairly solid. We do find that players tend to offer a bit less in the playoffs, which given the nature of playoff competition, is what we should expect. So next time you hear about a player raising the level of his game in basketball when it matters most, you might want to be a bit skeptical. Yes, a player can play better once in awhile. But to do this systematically seems unlikely.

As for the NFL, our data is too limited to draw any strong conclusions. And that goes for the conclusion that Grossman is a better quarterback in the playoffs than Manning. This might be true (I doubt it) but the data we have from this post-season is just not enough to tell this story.

– DJ

QB Score and Win Score are Discussed in the Following Post

Simple Models of Player Performance

Posted in: Football Stories