In modern football coverage, you will often hear analysts say a quarterback is "leading the league in EPA per play" or that a team's "defensive EPA has plummeted." If you are looking for a simple answer, EPA stands for Expected Points Added. It is a statistical metric used to measure the value of an individual play by calculating how much that specific play increased or decreased a team’s chances of scoring.

While traditional stats like passing yards or rushing touchdowns tell you what happened, EPA tells you how much it actually mattered in the context of the game. A 5-yard gain on 3rd-and-4 is a massive success, while a 5-yard gain on 3rd-and-15 is essentially a failure. EPA is the mathematical tool that allows us to quantify that difference.

The Foundation of Analytics: What Is Expected Points (EP)?

To understand Expected Points Added, you must first understand the concept of Expected Points (EP). This is the bedrock of modern football analytics.

Every single play in a football game occurs within a specific "game state." This state is defined by several critical variables:

  • Field Position: Where the ball is on the field (e.g., your own 20-yard line vs. the opponent's 10-yard line).
  • Down: Whether it is 1st, 2nd, 3rd, or 4th down.
  • Distance: How many yards are needed for a first down.
  • Time Remaining: How much time is left in the half or the game.

Data scientists have analyzed decades of NFL play-by-play data—hundreds of thousands of drives—to determine the average number of net points a team scores from any given situation. This average is the Expected Points.

For example, historical data shows that a team with a 1st-and-10 at their own 25-yard line (after a touchback) typically scores about 0.6 points on that drive. This doesn't mean they will score a fraction of a point on that specific drive; it means that across thousands of instances, the average outcome is 0.6 points.

Conversely, if a team has a 1st-and-Goal at the opponent's 1-yard line, their Expected Points value jumps to roughly 6.0. They are almost guaranteed a touchdown, but the possibility of a turnover or a settled field goal brings the average slightly below 7.

How Is EPA Calculated in the NFL?

The beauty of EPA lies in its simplicity. Once you have the Expected Points value for the start of a play and the end of a play, the calculation is a straightforward subtraction:

EPA = (Expected Points After the Play) – (Expected Points Before the Play)

Let’s look at a practical scenario to see this in action.

Scenario A: The High-Value Conversion

Imagine an offense is facing a 3rd-and-2 at the opponent's 40-yard line. At this moment, based on historical data, their EP might be 2.1. The quarterback completes a 5-yard pass, resulting in a 1st-and-10 at the 35-yard line. Because the team now has a fresh set of downs and is closer to the end zone, their new EP might be 2.8.

  • EPA Calculation: 2.8 (After) - 2.1 (Before) = +0.7 EPA

Scenario B: The "Empty" Yardage

Now imagine the same offense is facing a 3rd-and-25 at their own 20-yard line. Their EP is very low, perhaps -0.2, because they are likely to punt and give the ball back to the opponent in good field position. The quarterback completes a 10-yard pass. In traditional stats, this is a "10-yard completion." However, the team is now at 4th-and-15 and must punt. Their new EP might be -0.5.

  • EPA Calculation: -0.5 (After) - (-0.2) (Before) = -0.3 EPA

Even though the quarterback gained 10 yards, the play resulted in a negative EPA because the team’s situation actually worsened relative to their scoring potential. This is why EPA is the gold standard for separating "hollow" stats from winning plays.

Why Is EPA More Accurate Than Yards and Touchdowns?

Traditional statistics are often misleading because they lack situational awareness. In our experience analyzing NFL data, we have identified three primary reasons why EPA provides a clearer picture of player and team performance.

1. Eliminating the "Garbage Time" Bias

A quarterback might throw for 400 yards in a game, but if 250 of those yards came in the fourth quarter while his team was losing by 30 points, those yards are effectively meaningless. Defensive schemes change in "garbage time," often allowing short underneath routes to keep the clock moving.

EPA models often include filters for high-leverage situations. When we run our internal models, we frequently see "empty-stat" players drop significantly in efficiency rankings because their EPA on those late-game drives is near zero—the scoring probability didn't change because the game was already decided.

2. Rewarding Efficiency Over Volume

A running back who carries the ball 30 times for 100 yards (3.3 yards per carry) might look productive on paper. However, if most of those runs were 2-yard gains on 1st-and-10, he was constantly putting his team in "behind the sticks" situations (2nd-and-8).

EPA recognizes this. A 2-yard run on 1st-and-10 results in a negative EPA, whereas a 2-yard run on 3rd-and-1 results in a high positive EPA because it extends the drive. EPA rewards the player who does his job when the situation demands it, not just the player who accumulates volume.

3. Valuing the "Hidden" Plays

Sacks and turnovers are the most devastating plays for an offense, but they don't show up in passing yardage. An interception in the red zone can swing the EP from +4.5 to -1.5 (a 6-point swing). EPA captures this catastrophic loss perfectly, whereas a traditional box score might just show the quarterback had a "decent" day with 300 yards and one pick.

What Is a Good EPA per Play for a Quarterback?

In the analytics community, we don't just look at total EPA; we look at EPA per play. This levels the playing field for players in different offensive systems.

Based on data from the last several NFL seasons:

  • Elite Performance: +0.30 EPA/play or higher. This is the territory of MVP-caliber seasons. In our analysis of the 2022 and 2023 seasons, only a handful of quarterbacks reached this threshold consistently.
  • Strong Starter: +0.15 to +0.25 EPA/play. These are top-10 quarterbacks who efficiently move the chains and avoid costly mistakes.
  • League Average: +0.05 to +0.10 EPA/play. Because passing is generally more efficient than rushing, the average pass play usually has a slightly positive EPA.
  • Below Replacement Level: Negative EPA/play. If a quarterback is consistently in the negatives, they are actively hurting their team's chances of scoring every time they drop back to pass.

It is important to note that EPA is a "team-influenced" stat. While it is the best way to measure quarterback value, a quarterback’s EPA is also a reflection of his offensive line's protection and his receivers' ability to create separation. This is why analysts often pair EPA with CPOE (Completion Percentage Over Expected) to see how much of the efficiency is due to the quarterback's ball placement.

How to Interpret Defensive EPA

While EPA is usually discussed in terms of the offense, it is equally vital for evaluating defenses. For a defense, you want the lowest EPA possible.

When you look at a defensive leaderboard:

  • Negative Defensive EPA: This is excellent. It means the defense is consistently putting the opposing offense in worse situations (sacks, tackles for loss, turnovers).
  • EPA per Dropback Allowed: This is the most common way to measure pass defense. If a defense allows -0.10 EPA per dropback, they are shutting down the opponent's passing game.

Defensive EPA is often more "volatile" than offensive EPA. A defense might have a great EPA for three weeks because they faced backup quarterbacks, only to see it skyrocket when they face an elite signal-caller. This is why we look for sustained defensive EPA over a 4-to-8 week rolling average to determine if a unit is truly elite.

Success Rate: The Crucial Companion to EPA

EPA can sometimes be skewed by "explosive plays." For instance, an 80-yard touchdown pass creates a massive spike in EPA (+6.0 or more). A quarterback could have one lucky 80-yard bomb and 19 terrible plays, yet still have a decent "Average EPA" for the game.

To solve this, we use Success Rate.

A play is considered a "success" if the EPA is greater than zero. In simpler terms:

  • 1st Down: Was 40% of the yards needed gained?
  • 2nd Down: Was 60% of the remaining yards needed gained?
  • 3rd/4th Down: Was the first down or touchdown converted?

If a quarterback has a high EPA per play but a low Success Rate, they are "boom or bust"—relying on big plays but failing to move the chains consistently. The most dangerous offenses are those with both high EPA and a Success Rate above 50%.

What Are the Limitations of EPA?

As much as we rely on EPA for NFL analysis, it is not a perfect metric. There are several nuances that a raw EPA number might miss.

The Personnel Context

EPA doesn't know who is on the field. If a quarterback has a high EPA, is it because he is making elite throws, or because his wide receiver is catching a 5-yard slant and breaking three tackles for a 50-yard gain? While the "Expected Points After" will reflect the gain, the credit in a basic EPA model is often given entirely to the quarterback. Advanced "Air Yards EPA" models try to fix this by splitting the value between the throw and the run-after-catch.

The Weather Factor

Football is played in the elements. A game played in a blizzard or 40mph winds will naturally have lower EPA values because scoring is more difficult. When we evaluate players across different environments, we have to be careful not to penalize a "cold-weather" team for having lower EPA than a "dome" team.

The "Luck" Element

Fumble recoveries are essentially a 50/50 coin flip. However, recovering a fumble vs. losing it can result in a 4.0 EPA swing. A team that is "lucky" with fumble recoveries will have an inflated EPA that isn't necessarily sustainable or a reflection of their skill. This is why "Neutral EPA" (excluding turnovers) is sometimes used to find the true signal in the noise.

How Does EPA Influence Modern Coaching Strategy?

NFL coaches are increasingly using EPA and Expected Points models to make real-time decisions, most notably on 4th down.

In the past, coaches would punt on 4th-and-2 from the opponent's 45-yard line because of "field position." But the EP model tells a different story.

  • Punting: Gives the opponent the ball, resulting in a negative EP for you (e.g., -0.5).
  • Going for it: Even with a 50% success rate, the EP of a converted first down at the 43-yard line is so high (e.g., +2.5) that the "Expected Value" of going for it is significantly higher than punting.

This shift in philosophy, driven by EPA data, has led to the most aggressive era of football in history. Teams are going for it on fourth down more than ever because the math proves that the risk is worth the reward.

Summary: The Value of Expected Points Added

EPA has fundamentally changed how we watch and analyze football. It moves us past the era of "he passed for 300 yards" and into the era of "he added 15 points to his team’s score through his efficiency."

By accounting for down, distance, and field position, EPA provides a context-rich evaluation that traditional stats simply cannot match. Whether you are a fantasy football player looking for the next breakout star or a casual fan trying to understand why a 2-yard run felt so important, EPA is the key to unlocking the true narrative of the game.

FAQ: Common Questions About EPA in Football

What is a "good" EPA per play for a team? For an entire offense, anything above +0.10 EPA per play is considered very good. The best offenses in the league (like the recent Chiefs or 49ers) often hover between +0.15 and +0.20 for the season.

Does EPA include special teams? Yes, EPA can be calculated for kickoffs, punts, and field goals. A great punter who pins the opponent at the 1-yard line creates a high positive EPA for the special teams unit because he has significantly lowered the opponent's Expected Points.

Is EPA the same as Win Probability Added (WPA)? No. While they are related, WPA measures how much a play changed the team's chance of winning the game (0% to 100%). EPA measures the point value of the play. WPA is more sensitive to the clock and the current score—a touchdown when you are up by 30 points adds EPA but adds almost zero WPA.

Where can I find EPA stats for my favorite players? Several reputable analytics sites provide EPA data. Look for resources like nflfastR data repositories, Baldwin’s "RBSDM" (Rushing Backs Don't Matter) site, or the NFL’s Next Gen Stats portal for advanced situational breakdowns.

Why do some analysts hate EPA? Critics often argue that EPA is too reliant on historical data and doesn't account for individual "clutch" ability or the specific matchups of a game. While it’s true that EPA is a retrospective tool, its predictive power for future success is much higher than almost any traditional box score statistic.