Expected Points Added (EPA) Explained: The Metric That Changed Football Analytics

For decades, football analysis was dominated by a simple question: how many yards did that play gain? A 7-yard run was good. A 2-yard run was bad. A quarterback's value was measured by passing yards per game, and a running back's worth was captured in yards per carry. The entire framework for evaluating football was built on a unit of measurement -- the yard -- that treats every yard on the field as equally valuable.

But every football fan instinctively knows this is wrong. Gaining 5 yards on 3rd-and-4 is fundamentally different from gaining 5 yards on 3rd-and-15. The first converts a first down and keeps the drive alive. The second results in a punt. A 3-yard touchdown run is worth infinitely more than a 3-yard run on first-and-ten at your own 20. Yards do not capture what actually matters in football: scoring points and winning games.

Expected Points Added (EPA) solves this problem. It is arguably the single most important metric in modern football analytics, and it has fundamentally changed how teams, analysts, and informed fans evaluate players, plays, and strategies. Here is how it works.

What Are Expected Points?

Before you can understand EPA, you need to understand expected points (EP). Expected points represent the average number of points a team would be expected to score on its current drive, given the current down, distance, and field position.

The concept is built on historical data. By analyzing hundreds of thousands of real NFL plays, analysts can calculate the average scoring outcome from any game situation. For example:

Expected points also account for the possibility of the opposing team scoring. If you fumble or throw an interception, the other team gains possession in favorable field position. This negative scoring potential is baked into the expected points calculation, which is why situations deep in your own territory have low (or even negative) expected point values.

The expected points model creates a continuous value for every possible game state defined by three variables: down, distance to first down, and yard line. This transforms football from a sport measured in arbitrary yardage into one measured in its actual currency: points.

How EPA Is Calculated

Expected Points Added is simply the change in expected points from one play to the next. It answers the question: did this play make the team more or less likely to score, and by how much?

The formula is straightforward:

EPA = EP after the play - EP before the play

Consider a concrete example. A team faces 2nd-and-8 at their own 35-yard line. The expected points value of this situation is approximately 1.1. The quarterback throws a 15-yard completion, creating a new situation of 1st-and-10 at the 50-yard line, which has an expected points value of approximately 2.0.

The EPA of that play is 2.0 - 1.1 = +0.9 EPA. The play added nearly a full expected point to the team's scoring potential.

Now consider a different outcome. Same starting situation -- 2nd-and-8 at the own 35 -- but the quarterback throws an interception, and the opposing team takes over at the 35. The expected points value flips: the original team now has an expected scoring value of roughly -3.5 (because the opponent is likely to score from that field position).

The EPA of the interception is -3.5 - 1.1 = -4.6 EPA. The play destroyed nearly five expected points. This captures something that traditional stats miss entirely: a turnover deep in your own territory is catastrophic, far worse than a turnover near midfield.

Why Yards Are Misleading

Traditional yardage statistics are seductive because they are simple. But simplicity is a liability when it obscures reality. Here are several ways yards mislead:

Yards ignore context. A 4-yard gain on 3rd-and-3 is far more valuable than a 4-yard gain on 3rd-and-10. The first creates a new set of downs; the second ends the drive. Yards treat them identically. EPA distinguishes them clearly.

Yards ignore field position. Gaining 20 yards from your own 5 to your own 25 is far less valuable than gaining 20 yards from the opponent's 25 to the opponent's 5. The second gain virtually guarantees points; the first merely moves you out of dangerous territory. EPA captures this difference because expected points are calibrated to field position.

Yards ignore turnovers properly. A quarterback who throws for 300 yards but has two interceptions might look productive in the box score. But EPA reveals that the interceptions likely destroyed more value than the completions created. A quarterback with 220 yards and zero turnovers may have contributed more to scoring by EPA.

Yards ignore the value of scoring. A 1-yard touchdown run is worth 1 yard in traditional stats but is enormously valuable in EPA terms because it converts field position into actual points. Similarly, a 50-yard completion that ends at the 1-yard line is worth much more in EPA than a 50-yard completion from your own 20 to your opponent's 30, because the former virtually guarantees a score.

The fundamental insight is that not all yards are created equal. EPA weights each play by its actual impact on scoring, which is what ultimately determines wins and losses.

EPA Per Play: Comparing Players and Teams

Raw EPA totals are useful but can be misleading because they do not account for volume. A quarterback who plays all 17 games will accumulate more total EPA than one who plays 12, even if the latter is more efficient on a per-play basis.

EPA per play (sometimes written as EPA/play) controls for volume and is the standard efficiency metric in modern football analytics. It answers the question: on an average play, how much does this player or team increase (or decrease) their expected scoring?

For quarterbacks, EPA per dropback (EPA per pass attempt, including sacks) is the gold standard efficiency metric. Here is what the scale generally looks like in the NFL:

These numbers might seem small, but they compound over hundreds of plays. A quarterback who averages +0.20 EPA per dropback over 600 dropbacks adds roughly 120 expected points over the season compared to a replacement-level player -- a massive advantage.

EPA vs. Traditional Stats for Evaluating Quarterbacks

The quarterback position is where EPA's superiority over traditional stats is most evident. Consider how traditional stats can mislead:

Passer rating (the NFL's official quarterback rating system) does not account for sacks, does not weight situations by context, and uses an arcane formula that caps individual components at arbitrary thresholds. A quarterback who compiles a high passer rating on safe, short throws in garbage time will look better than a quarterback who takes calculated risks in critical moments.

Passing yards reward volume regardless of efficiency. A quarterback who throws 50 times in a game because his team is trailing will accumulate yards without necessarily playing well. Conversely, a quarterback whose team builds an early lead and then runs the ball will have fewer yards despite having played brilliantly.

Touchdowns and interceptions are important but lack context. A touchdown on a 1-yard play after a long drive is less attributable to the quarterback than a 40-yard touchdown pass. An interception thrown on a desperation heave at the end of a half is far less damaging than an interception thrown in the opponent's territory on a routine play.

EPA incorporates all of this context. It weights every throw by the situation -- down, distance, field position, and scoring -- and produces a single, coherent measure of a quarterback's contribution to his team's scoring. When you see EPA-based quarterback rankings, they often differ significantly from traditional stat-based rankings, and they tend to be more predictive of future performance and team success.

Visual Examples: Seeing EPA in Action

One of EPA's strengths is that it lends itself to powerful visualizations. Two of the most common are drive charts and scatter plots.

EPA drive charts show the expected points value at each point during a drive, creating a line that rises and falls with each play. A successful drive looks like a steadily climbing line from low EP to a touchdown (worth approximately 7 points). A drive that stalls and ends in a punt shows a line that rises slightly, plateaus, and then drops. A drive that ends in a turnover shows a dramatic plunge. These charts make it visually obvious which plays were the most impactful -- the steepest rises and drops.

EPA scatter plots are commonly used to compare team or player performance in passing and rushing. A typical team-level scatter plot shows EPA per pass play on one axis and EPA per rush play on the other. Teams in the upper-right quadrant are efficient at both; teams in the lower-left are inefficient at both. This immediately reveals team profiles in a way that raw yardage totals cannot.

Cumulative EPA charts track a player's total EPA over the course of a season, game by game. These charts reveal consistency (a steady upward slope) versus volatility (sharp peaks and valleys) and allow easy comparison between players.

How Teams and Analysts Use EPA

EPA is not just an academic exercise. It has practical applications throughout professional football.

Play-calling decisions. EPA data has been instrumental in the analytical revolution around fourth-down decision-making. Traditional coaching wisdom said to punt on fourth down in almost all situations. EPA analysis shows that going for it on 4th-and-short in the opponent's territory is often the higher-expected-value decision, because the expected points gained from converting outweigh the expected points lost from failing. Coaches who adopt aggressive fourth-down strategies -- most notably, former Eagles coach Doug Pederson and current Lions coach Dan Campbell -- tend to accumulate EPA advantages over the course of a season.

Personnel evaluation. Teams use EPA to evaluate free agents, draft prospects (using college EPA data), and current roster players. Because EPA controls for context, it can distinguish between a running back who gains yards because of his offensive line and one who creates yards independently.

Game planning. Defensive coordinators use EPA to identify an opponent's most and least efficient play types, personnel groupings, and tendencies. If an offense has a high EPA on play-action passes and a low EPA on traditional dropbacks, the defense can scheme accordingly.

Broadcasting and media. EPA has become increasingly visible in football media. Analysts on networks and podcasts regularly cite EPA to support arguments, and EPA-based graphics appear during broadcasts, helping fans understand the significance of individual plays in real time.

The Limitations of EPA

No metric is perfect, and EPA has important limitations that responsible analysts acknowledge.

EPA does not assign individual credit. A quarterback's EPA is influenced by his receivers, offensive line, play-calling, and coaching. Disentangling individual contributions from team context remains one of the hardest problems in football analytics. Metrics like CPOE (Completion Percentage Over Expected) attempt to isolate the quarterback's contribution, and they are often used alongside EPA for a more complete picture.

EPA models are built on averages. The expected points values represent historical averages, which means they may not perfectly capture the specifics of any individual game. A team with an elite defense may score fewer points from a given field position than the model predicts, and vice versa.

EPA does not account for game situation equally well. Late-game situations -- when teams are protecting leads, attempting comebacks, or managing the clock -- involve strategic considerations that EP models based on average situations may not fully capture. Win Probability Added (WPA) is a complementary metric that better handles these late-game dynamics.

Despite these limitations, EPA represents an enormous improvement over traditional statistics. It transforms football analysis from a sport measured in an arbitrary unit (yards) to one measured in the unit that actually matters (points), and it provides a common language for evaluating every play, player, and strategy against a single standard.

The next time you watch a football game, pay attention not just to how many yards a play gained but to the situation in which it occurred. That 3-yard run on 3rd-and-2 was worth more than the 8-yard pass on 1st-and-10. EPA knows the difference, even if the box score does not.

Read our free NFL Football Analytics and College Football Analytics textbooks for the full deep dive.