EPA, or expected points added, measures how much a single play changed a team's expected score. Every down, distance and field position has an expected point value based on 25 years of NFL drives; a play's EPA is the difference between that number before the snap and after the whistle. A 5-yard run on third-and-4 has positive EPA because it kept the drive alive. A 5-yard run on third-and-8 has negative EPA because it did not.
Updated Sept. 16, 2026.
Week 1's lowest-EPA game was Atlanta Falcons at Pittsburgh Steelers: -36.8 total EPA across both offenses, the 194th-lowest of 7,289 games since 1999. The best was Chicago Bears at Carolina Panthers at +39.9. The next game down the all-time list is Baltimore Ravens at Cincinnati Bengals in 2007 (-36.9).
Total EPA is the sum of expected points added on every pass and run by both offenses. Negative means the offenses, between them, lost expected points over the game.
| Game | Total EPA | Percentile since 1999 |
|---|---|---|
| Atlanta Falcons at Pittsburgh Steelers | -36.8 | 3 |
| Miami Dolphins at Las Vegas Raiders | -14.3 | 24 |
| Green Bay Packers at Minnesota Vikings | -12.9 | 27 |
| Tampa Bay Buccaneers at Cincinnati Bengals | -9.5 | 34 |
| Denver Broncos at Kansas City Chiefs | -5.3 | 42 |
| New England Patriots at Seattle Seahawks | -4.3 | 45 |
| San Francisco 49ers at Los Angeles Rams | -3.4 | 47 |
| New Orleans Saints at Detroit Lions | -2.6 | 48 |
| Arizona Cardinals at Los Angeles Chargers | -2.5 | 49 |
| New York Jets at Tennessee Titans | +4.3 | 63 |
| Baltimore Ravens at Indianapolis Colts | +6.5 | 67 |
| Washington Commanders at Philadelphia Eagles | +7.7 | 70 |
| Cleveland Browns at Jacksonville Jaguars | +8.9 | 72 |
| Buffalo Bills at Houston Texans | +17.0 | 85 |
| Dallas Cowboys at New York Giants | +35.5 | 98 |
| Chicago Bears at Carolina Panthers | +39.9 | 99 |
Yards treat every yard the same. EPA does not. Before each play, a model looks at down, distance, yard line, time left and the score and asks: from here, how many points does the offense usually end up with on this drive, minus what the other team gets? That number is the expected points. First-and-10 at your own 25 is worth about +0.9; first-and-goal at the 2 is worth almost 6; fourth-and-15 at your own 10 is negative, because the likeliest outcome is a punt that sets the other team up.
EPA is simply the change. A touchdown from the 2 adds about 1 point of EPA (it was already worth nearly 6). A 60-yard bomb from your own 20 adds around 4. An interception returned for a score can cost more than 10, because the offense went from expecting points to giving them up.
Summed over a game, a season or a career, EPA turns every play into the same currency: points. That is why it replaced yards and passer rating as the number analysts reach for first. It rewards what actually scores, punishes what actually loses, and it is the backbone of most public win-probability models.
A quarterback throws 300 yards with two long touchdowns in a blowout win and finishes at +18 EPA. Another throws 300 yards, but 120 of them come in garbage time on drives that end in field-goal attempts, plus one interception in the red zone, and finishes at -2 EPA. Same yards, opposite games. The box score cannot tell them apart; EPA can.
EPA = expected points after the play minus expected points before the play. Expected points come from a model (nflfastR's is the public standard) fit on every play since 1999, using down, distance, yard line, half, time remaining, timeouts and score differential. Turnovers flip the sign of the next expected value because possession changes hands.
For a team offense over a season, +0.10 EPA per play is elite, 0 is average, and -0.10 is among the worst in the league. For a quarterback, +0.20 per dropback is an MVP-level season; the league average dropback is close to +0.05 because passing is more efficient than running.
Because a yard is worth different amounts in different spots. Three yards on third-and-2 wins the down; three yards on third-and-9 loses it. EPA prices each yard by what it did to the expected score, so it credits the plays that actually move a team toward points.
More than most. Offensive EPA per play is one of the stickier team stats from one half of a season to the next, and it predicts point differential better than yards or turnover margin. Defensive EPA is noisier and regresses harder.
Total EPA adds up every play for both offenses. A clean, well-played game lands well above zero. A game full of punts, sacks and turnovers lands below it, which is why the worst games by total EPA are the ones nobody wants to rewatch.
Category: Stats & Analytics. Part of the StickToTheModel NFL Encyclopedia.