What the heck is DVOA? A primer for more advanced football statistics

If you follow baseball more than casually, you’ve probably heard of “sabermetrics”, the use of statistics to analyze some of baseball’s long-held questions. Well, it’s taken hold in football, too, with all its acronym-ridden glory. One of the pioneers of the new brand of football statistics is Football Outsiders, whose numbers I will be using for most of my articles. A summary and explanation of their best statistics follows. They explain it (probably better) here, but I’ll be breaking it down as simply as I can.

Their main statistic (the one that gets all the press coverage) is DVOA, or Defense-adjusted Value Over Average. The beauty of this statistic is that it can be used to compare teams as a whole or in any situation you can think of as well as players. Third downs in the fourth quarter? Yep. Second-and-8? Definitely.

Here’s an example of how DVOA is calculated: Suppose the Steelers are facing the Bengals, have a third-and-5 on their own 35, and Rashard Mendenhall rushes for six yards. You place a value on that six yard run, compare it to how other third-and-5’s fare around the league, and adjust for the quality of the Bengals defense.

Likewise, the same play is scored for the Bengals defense, comparing the result to how other defenses fare in that situation, then adjusting for the quality of the Steelers offense. It’s still called DVOA, though, because they don’t want to inundate you with acronyms. They’d be crazy to do that!

As an example, the team DVOA rankings for 2008 are here. As you can see, the teams’ offensive, defensive, and special teams ratings are added together to make the complete ranking, and that teams are positioned around 0. So that year, the Steelers were 29.6 percent better than an average team, and the Lions were 48.4 percent worse.

To sum up, here’s what DVOA can do:

  • Compare teams overall and their offenses and defenses with better accuracy than standard measures such as yards gained and yards allowed, which don’t adjust for opponent quality.
  • Provides one way of comparing individual offensive players.
  • Combines all aspects of special teams (field goals, kick and punt returns and coverage) into one rating.
  • During the season, provides a good way to calculate playoff odds.

Here’s what it can’t do:

  • It can’t compare players and teams across history, as the detailed and complete play-by-play data is only available back to 1994.
  • It also can’t value individual defensive players, as things like dropped interceptions (think Ike Taylor, though I don’t have the data to know whether or not that’s just my perception) are not included in the play-by-play they do have.
  • It isn’t able to separate a player’s performance from those of his teammates on (or off) the field, which means really interesting questions like “How will Willie Parker fare on the Redskins?” and “How would the Steelers perform differently with a different playcalling scheme?” cannot be fully explored at this moment.

Defense-adjusted Yards Above Replacement, or DYAR, is the same basic DVOA calculation and adjustments, but rescaled to allow for comparison of individual players. However, they are compared not to the average player, but to a replacement-level player, which is the level of performance a typical backup would produce. Sometimes, though, the backups are better. In 2008, for example, Mewelde Moore had 139 DYAR to Parker’s 46. Moore’s performance that season ranked 14th in the NFL.

If you have any questions about these and other stats you’ll find on those pages, leave a comment and I’ll respond. From now until game 1, I’ll be using DVOA and DYAR to analyze the Steelers and the other teams in the AFC Central.