Public discussion of football analytics has exploded in recent months. The NFL is expanding what we can do with analytics thanks to player tracking, and it is making some of that player tracking available publicly with events such as the Big Data Bowl. Numerous teams have expanded their analytics departments. Lots of groundbreaking work is being done both by large analytics companies and by independent analysts simply posting their results on Twitter.
Despite this recent upsurge, many of the findings of the analytics community haven't yet filtered out to the larger football community. It's not just regular fans; even national commentators will still talk about momentum and establishing the run and other ideas that have been generally disproved. If you want to be a smarter fan, you need to look at the game as it's played today, not as it was played 20 or 30 years ago. And you should know what the analytics is telling us ... in part because the smartest organizations know these things as well.
Let's look at some of the NFL myths that have been busted by analytics. Some of these findings go all the way back to the first public writing about football analytics, with the publication of "The Hidden Game of Football" back in 1988. Other findings are more recent. But these are all ideas you should be skeptical about if you hear them coming from the mouth of your favorite football commentator this season.
"Sacks are the best measure of a pass rush"
Pittsburgh and Kansas City led the NFL with 52 sacks apiece on defense last season. Chicago and Minnesota were next, with 50. So we can expect each of these teams to have a strong pass rush in 2019, right?
Not necessarily. Sack totals from one season don't necessarily help us predict sack totals the next. In fact, even within the same season, sack totals aren't very good for telling us how good a pass rush will be in the next few games.
For example, in the first half of last season, the top pass rushes in terms of sack rate were Minnesota, Detroit and Green Bay. In the second half of the season, those teams ranked seventh, 21st and 28th, respectively. The top pass rush of the second half by sack rate was New Orleans. In the first half of the season, the Saints had ranked 26th.
Overall, for the past three seasons, the correlation between sack rate in the first half of the season and sack rate in the second half of the season was a tiny 0.09. Compare that to pressure rate, measuring sacks plus hurries charted separately from the standard play-by-play. The correlation between pressure rate in the first half of the season and pressure rate in the second half of the season was a much more useful 0.43.
Both sacks and pressures are more predictive on the offensive side of the ball, but once again, pressure rate is the more predictive stat. Over the past three seasons, the first half-second half correlation of sack rate was 0.37. The correlation of pressure rate was 0.55.
The problem, of course, is that pressures are an unofficial stat. The rates I used here feature data calculated by FO's partners at Sports Info Solutions, but different pressure rates and stats are tabulated by different companies, from ESPN Stats & Information to Pro Football Focus to Sportradar. And how each company computes pressure rate is different as well: Some pressure rates incorporate quarterback knockdowns even if the knockdown came well after the pass and didn't affect the quarterback's throwing motion. Nonetheless, any measure of pressure rate is probably a better guide to both pass rush and pass protection than looking just at sack totals.