Defense-adjusted Value Over Average — a play-by-play efficiency rating relative to league baseline.
Defense-adjusted Value Over Average was developed by Aaron Schatz at Football Outsiders in the early 2000s and remains one of the longest-running play-by-play team metrics. Like EPA, it scores every play against a league-average benchmark. Unlike EPA, it does so on the success-value scale rather than the expected-points scale, and it explicitly adjusts each play for the quality of the opponent.
Each play is scored against a situational baseline — what an average play in that down, distance, and field position is worth — and that baseline value is subtracted from the actual play value. The differences sum to a team's raw value over average. Opponent adjustments then scale every play by the strength of the defense (or offense) the team faced. A 20-yard completion against the league's best secondary counts for more than the same play against the worst.
DVOA is expressed as a percentage above or below average. A 20% offensive DVOA means a team was 20% more productive per play than league average against an average opponent; -10% means 10% below. The metric is split into offensive, defensive, and special-teams components, and further into rushing, passing, and situational splits.
Compared with raw EPA, DVOA's strength is the explicit opponent adjustment, which makes its ratings comparable across teams that played different schedules. Its weakness is that the success-value scale is proprietary and harder to inspect than the cleanly empirical expected-points scale. Most public modeling has migrated to EPA-based ratings, but DVOA remains a useful cross-check, especially for teams with very strong or weak schedules.
We use DVOA as a sanity check against our EPA-based ratings. When the two disagree by more than two standard deviations, we re-examine the underlying play-by-play before publishing a pick.