What Is Success Rate in Football? Why It Can Disagree With EPA and Fantasy Points

Success rate measures how often plays satisfy a stated definition of success. Under the EPA method explained here, a play succeeds when it produces positive expected points added. The rate tells you the frequency of those positive results, not how large they were. That is why an offense can win the success rate comparison while losing the EPA per play comparison.

Information cutoff: October 19, 2026 at 4:32 AM Central. This is a statistics guide, not current NFL rankings. All play samples, EPA values and fantasy lines below are hypothetical Football Nation teaching examples. The EPA values are stipulated for arithmetic, not claimed outputs from a live model. Recommendations are FN editorial analysis.

State the definition before quoting the percentage

nflfastR’s clean_pbp documentation defines its success indicator using EPA greater than zero. Its source code marks positive EPA as one and other observed EPA values as zero. An EPA of exactly zero therefore does not count as a success in this definition. Missing EPA remains missing, rather than being silently counted as failure.

For a declared sample with observed EPA, divide successful plays by qualifying plays and multiply by 100. Eighteen successes on 40 plays is 45%. The label should identify the sample: for example, offensive success rate across the selected observed plays, or rushing success rate across a specified rushing subset.

For the underlying scoring concept, see our EPA guide. This article focuses on turning individual play values into a frequency and interpreting what that frequency leaves out.

Why another website may report a different success rate

Not every success rate uses EPA. FTN’s separate running back definition uses yardage thresholds by down, with specified fourth quarter adjustments. Its documentation also distinguishes that measure from the success framework used for DVOA. Do not mix those labels or assume every percentage comes from the same test.

Named approach	What determines success
EPA success used in this guide	A qualifying play has EPA greater than zero
FTN running back success rate	A rush meets its published down and distance yardage benchmark, including stated exceptions
Third down conversion rate	The selected third down produces a conversion; this is a different denominator and question

A comparison must use the same definition on both sides. Also verify whether a defensive rate describes successes allowed to the offense or successes credited to the defense. A lower offensive success rate allowed is a better defensive result under the first convention. The column heading needs that perspective.

Example one: frequency is not magnitude

Imagine two five play samples. Sample A has four plays worth plus 0.10 EPA apiece and one worth minus 2.00. Sample B has four plays worth minus 0.10 apiece and one worth plus 3.00. Every play qualifies and has an observed value.

Hypothetical result	Sample A	Sample B
Successful plays	4 of 5	1 of 5
Success rate	80%	20%
Total EPA	0.40 minus 2.00 = minus 1.60	3.00 minus 0.40 = plus 2.60
EPA per play	Minus 0.32	Plus 0.52

A produces positive value more often, but its single large loss overwhelms its small gains. B fails the positive value test more often, but its large gain outweighs those losses. Neither summary is mathematically wrong. They emphasize different features of the same stipulated results.

This is not proof that either pattern is stable, desirable or repeatable. Five plays are a demonstration, not a team identity. Use success rate beside EPA magnitude and the actual play descriptions before deciding what the offense needs to fix.

Example two: weight the underlying opportunities

Suppose an offense succeeds on 12 of 20 qualifying plays in one game, a 60% rate, and 32 of 80 in another, a 40% rate. Its combined record is 44 successes across 100 plays, or 44%. The simple average of the two percentages is 50%, which answers a different question by weighting each game equally.

Neither method should be disguised as the other. If your subject is the combined set of plays, pool the successes and opportunities. If a chart deliberately averages games, say so. A short appearance, a long game and a filtered subset can otherwise create a misleading comparison.

What success rate does not promise a fantasy manager

It does not award fantasy points or tell you a player’s share of the offense. Consider a fictional receiver with seven catches for 42 yards and no other statistics. Under specified scoring of 0.1 point per receiving yard, he earns 4.2 points in standard, 7.7 in half PPR and 11.2 in full PPR.

A second receiver with one catch for 30 yards earns 3.0, 3.5 and 4.0 in those same formats. These fantasy calculations do not require either player’s targets to have positive EPA. You cannot infer the success rate from the fantasy score, or reconstruct the fantasy score from the success rate.

Our PPR explainer details the reception bonus. For opportunity, use route participation and targets. A reserve in a successful offense may still be a poor lineup choice if the relevant work belongs to someone else.

How Football Nation would apply the evidence

When evaluating an offense, ask whether frequent failures arise in a particular part of the game: early downs, protection situations or a defined personnel group. Then examine the actual plays and opponent. A broad percentage cannot isolate the blocker, passer, receiver or coach responsible.

For a waiver or trade decision, require a player specific explanation. Did his opportunities change, or did the team merely post a different efficiency result? Does the asking price assume that every teammate benefits? Our points per game comparison guide adds the availability and sample questions.

The strongest counterargument is that explosive plays can decide games, making frequency less important than total value. The teaching example supports that caution. Success rate is useful precisely as a complementary measure: it helps you notice whether the offense’s result came from many helpful plays or a few large events.

Common questions

Does a successful play always gain a first down? Not under the EPA definition. It needs positive modeled scoring value. Our first down guide covers the separate chain moving concept.

Does an unsuccessful play prove a bad decision? No. A result and the choice that produced it are different evaluations. Consider alternatives, uncertainty and the team’s objective before judging the call.

Is this rate automatically adjusted for opponents? No. Do not assume an adjustment unless the named provider and measure explicitly include it.

Football Nation conclusion

Write the definition beside the number. Count the right plays, compare frequency with magnitude and then return to the individual workload. The next checkpoint is whether the same explanation holds in a larger comparable sample, not whether one good game makes a percentage look impressive.

Sources checked October 19, 2026 at 4:32 AM Central: nflfastR success documentation, its calculation source, the running back section of FTN’s methodology explainer and ESPN Scoring Formats. No provider leaderboard values are reproduced.