Understanding EPA Tiers

What is Expected Points Added (EPA)?

Expected Points Added measures the value of each play by calculating how much a team's actions increase or decrease their likelihood of scoring points on a particular possession. Unlike traditional stats, EPA accounts for contextual factors like field position, down & distance, and time remaining.

New to EPA? Read our primer, What is Expected Points Added?.

How EPA Tiers Work

Since EPA can be measured at an individual play level, it can be aggregated across all of the same splits used for other stats. One of the most commonly used splits for EPA is offensive and defensive EPA per play.

EPA tiers represent this data in a scatter chart by plotting teams by their offensive (x-axis) and defensive (y-axis, inverted). Teams with better offenses will be out to the right, and teams with better defenses will be high vertically.

Teams that are good at both will be in the upper right quadrant, meaning you can think of the best teams as those being furthest "up and to the right."

Additionally, since net EPA per play is the difference between offensive and defensive EPA, teams along the same left to right diagonal line can be thought of as comparable, even if their contributions to net EPA are split differently between offense and defense.

Because offensive performance is more predictive than defensive performance, the best "diagonal" to use when comparing teams is actually one with a steeper slope (1.6). A team with slightly worse net EPA might be in a higher tier if their offense is the primary driver of their success.

For a deeper exploration of team strength and for a comparison to other ranking models like Elo, check out our ELO-based power rankings.

Other Ways to Use Expected Points Added

Team offense and defense are not the only way to aggregate EPA. It can also be used as a directional measure for individual players, by averaging EPA on plays in which they were involved.

For instance, with quarterbacks, we typically look at EPA per dropback, while for receivers, we look at EPA per target or reception. You can explore these individual EPA metrics in depth in our QB EPA Rankings and WR EPA Rankings.

Frequently Asked Questions

What is a good EPA per play?

League average EPA per play is meant to be 0.0 by definition. Positive EPA indicates above-average performance, while negative EPA suggests below-average play. Elite offenses typically sustain +0.10 to +0.20 EPA per play.

Since defensive EPA is more dependent on opponent, any negative EPA is considered quite good.

How does EPA compare to DVOA?

Both EPA and DVOA measure efficiency, but EPA uses expected points while DVOA uses yards and success rate. EPA tends to correlate more strongly with actual scoring and game outcomes because it is modeled directly on points scored and allowed.

Why do offensive tiers have more weight?

Historical analysis shows offensive EPA per play is stickier across different windows, which speaks to a greater predictive power than defensive EPA. Using a weight of 1.6 for offensive EPA and 1.0 for defensive EPA tends to be most optimal for predicting future net EPA.

What's the difference between EPA tiers and power rankings?

EPA tiers focus specifically on Expected Points Added as a measure of observed efficiency. While this is a good starting point for predictive rankings, models that consider additional factors like home field advantage, opponent quality, and pre-season rankings like nfelo tend to perform better. nfelo's rankings are available on our NFL Power Rankings page.