// REFERENCE
ADVANCED STATS GLOSSARY
Every metric used on this site, defined in one place. If you've seen a label in the tools and wondered what it means, it's here. If it's not here, it shouldn't be in the tools.
Data sources and limitations are included. These aren't decorative footnotes — they matter. A metric from a public play-by-play model and a metric from the NFL's proprietary tracking system are different things, even when they share a name.
// SECTIONS
Efficiency Metrics (EPA, WP)
Quarterback / Passing Metrics
Running Back / Rushing Metrics
Receiver / Receiving Metrics
General Game Metrics
// EFFICIENCY METRICS
Expected Points Added
EPA
The change in expected points produced by a single play. Calculated as: EPA = Expected Points After Play − Expected Points Before Play.
Positive EPA means the play advanced the offense (generated expected value); negative EPA means it hurt.
EPA accounts for down, distance, field position, and historical point outcomes from similar situations.
See the full EPA explainer.
Source: nflverse open-source expected points model (nflfastR)
EPA per Play
EPA/PLAY
A player's or team's average EPA across all plays they were involved in. Better than yards per play because it's situation-adjusted.
One of the most predictive season-to-season efficiency metrics in football analytics.
Positive values mean above average; negative values mean below. League average hovers near 0.
Source: nflverse play-by-play
EPA per Drive
EPA_TOTAL
Total EPA across all plays in a drive. Shown in the Drive Tracker as EPA_TOTAL.
A drive that gains 40 yards but punts from midfield often has negative EPA.
A 5-play TD from the 20 can have +3.5 EPA. The Drive Tracker uses this to rank drives by impact.
Source: nflverse play-by-play, aggregated per drive
Win Probability
WP
The probability (0–100%) that the home team wins, given current score, time remaining, field position, and down/distance.
Updates after every play. Not a talent model — a situational model based on historical outcomes from identical game states.
See the full Win Probability explainer.
Source: nflverse WP model (logistic regression on historical play-by-play)
Note: does not account for team quality, QB skill, or injuries — situational only.
Win Probability Swing
WP_SWING
How much a drive (or play) moved the Win Probability needle. Drives with large positive WP swings are the ones that
actually determined the game's outcome. A 7-play 80-yard TD drive in the fourth quarter of a close game
can swing WP by 30–40 percentage points — that's the drive that won the game.
Source: nflverse WP model, aggregated per drive
// QUARTERBACK / PASSING METRICS
Completion Percentage Over Expected
CPOE
Actual completion rate minus expected completion rate, where expected completion is modeled from throw difficulty
(air yards, receiver separation, pressure, pass location). Positive CPOE means the QB is outperforming expectations;
negative means underperforming. Adjusts for the fact that short, high-percentage offenses inflate raw completion rate.
See the full CPOE explainer.
Source: Next Gen Stats (NGS) — official NFL player tracking data
Passer Rating
PASSER_RTG
The traditional NFL passer rating formula: combines completion %, yards per attempt, TD rate, and INT rate
into a 0–158.3 scale. Widely used; widely criticized for omitting sacks and not adjusting for situation.
Shown for reference alongside CPOE and EPA-based metrics. League average is roughly 90.
Source: nflverse play-by-play
Note: use alongside EPA/play; passer rating alone is insufficient for modern QB evaluation.
Time to Throw
TIME_TO_THROW
Average time in seconds from the snap to the QB's throw. Lower TTT means quick decisions and processing;
higher TTT means holding the ball longer (either taking shots downfield or under pressure).
Elite pocket passers typically fall in the 2.4–2.8 second range. Above 3.0 seconds often correlates with
pressure or scrambling.
Source: Next Gen Stats (NGS)
Aggressiveness
AGGRESSIVENESS
The percentage of a QB's throws that traveled to "tight windows" — areas where the nearest defender
was within 1 yard of the receiver. High aggressiveness means the QB is taking chances and targeting
covered receivers; low aggressiveness means throwing to open targets. Neither is universally good:
context depends on scheme and down/distance.
Source: Next Gen Stats (NGS)
Max Air Distance
MAX_AIR_DIST
The maximum distance (in yards) a QB threw the ball downfield from the line of scrimmage in a season.
A measure of arm strength and willingness to take deep shots. Shown for context alongside aggressiveness.
Source: Next Gen Stats (NGS)
Average Depth of Target
ADOT
Average air yards (depth downfield) of all pass attempts by a QB or to a receiver. High ADOT = deep attack;
low ADOT = short passing game. Context matters: a WR with a 12-yard ADOT who catches 65% of targets
is more impressive than one with a 3-yard ADOT at 65%.
Source: Next Gen Stats (NGS) / nflverse air yards data
// RUNNING BACK / RUSHING METRICS
Rushing Yards Over Expected
RYOE
Total rushing yards minus expected rushing yards, where expected is modeled from the run situation:
defenders in the box, offensive formation, down/distance, field position, and pre-snap alignment.
A runner who generates positive RYOE is outperforming what the situation should yield.
A runner generating negative RYOE may be running into stacked boxes or failing to exploit created space.
Source: Next Gen Stats (NGS)
RYOE per Attempt
RYOE/ATT
RYOE divided by number of rush attempts. The per-carry version of RYOE — normalizes for volume
and lets you compare RBs who carried the ball different numbers of times.
Positive = outperforming model expectations on a per-carry basis.
Source: Next Gen Stats (NGS)
Efficiency
EFFICIENCY
The percentage of a RB's rush attempts that gained positive yardage (8-in-box or not).
A rough measure of how often the runner does something with the carry,
rather than being stopped at or behind the line.
Source: Next Gen Stats (NGS)
Time to Line of Scrimmage
TLOS
Average time (in seconds) from snap to when the ball carrier reaches the line of scrimmage.
Lower TLOS means the RB is hitting the hole quickly; higher values suggest hesitation or blocked
runs that require cutback or patience. Compare with RYOE/att for context.
Source: Next Gen Stats (NGS)
8 in Box Percentage
8_IN_BOX%
The percentage of rush attempts a RB faced with 8 or more defenders in the box pre-snap.
High 8-in-box% means the defense was loading up against the run; generating positive RYOE against
those looks is more impressive than against lighter boxes.
Source: Next Gen Stats (NGS)
// RECEIVER / RECEIVING METRICS
Separation
SEPARATION
Average distance (in yards) from the nearest defender at the catch point.
High separation = consistently getting open. Influenced by route running, quickness, and scheme —
not purely an individual skill in isolation, but correlates strongly with consistent receiving production.
Source: Next Gen Stats (NGS)
YAC Over Expectation
YAC+/−
Yards after catch compared to what is expected based on catch location and field position.
Positive YAC+/− means the receiver consistently does more with the ball after the catch
than receivers typically do from the same spot. A measure of run-after-catch ability,
separate from route running or separation.
Source: Next Gen Stats (NGS)
Average Intended Air Yards
AVG_IAY
Average depth of the QB's aim point (where the ball was intended to go) per target, regardless of completion.
Different from ADOT (which measures where the ball traveled) — IAY measures the intended throw depth,
including incomplete passes. Higher IAY means deeper routes; lower means shorter targets.
Source: Next Gen Stats (NGS)
Air Yards Share
AIR_YDS_SHARE%
The percentage of the team's total air yards on completed passes that went to this receiver.
A high air yards share means the receiver is the primary deep target or sees significant volume
on longer routes. A low share doesn't mean the receiver is unimportant —
many high-value slot receivers accumulate yards after catch rather than in the air.
Source: nflverse play-by-play / air yards data
Cushion
CUSHION
The distance (in yards) between the receiver and the nearest defender at the snap.
High cushion means the corner is playing off-coverage (anticipating a deep route or protecting against speed).
Low cushion means press coverage. Context matters: a receiver who beats press coverage has
earned every yard; one who always sees soft cushion should have better raw separation numbers.
Source: Next Gen Stats (NGS)
// GENERAL GAME METRICS
Tempo
TEMPO (SEC/PLAY)
Average seconds between plays during a drive. Lower tempo (hurry-up) can tire defenses and
limit substitutions; higher tempo suggests clock-control or deliberate pace.
In the Drive Tracker, TEMPO is shown per drive.
Source: nflverse play-by-play, computed per drive
Turnover
TO / OPP_TO
Fumbles lost + interceptions. TO = turnovers committed by the selected team.
OPP_TO = turnovers committed by the opponent (i.e., turnovers forced).
Turnovers produce the largest single-play EPA swings in football — they flip expected
points from one team to the other.
Source: nflverse play-by-play
// NOTE ON DATA SOURCES
Metrics sourced from nflverse / nflfastR are open-source and reproducible. Metrics sourced from Next Gen Stats (NGS) use the NFL's proprietary player-tracking infrastructure — optical tracking of every player on every play. NGS data is licensed to the NFL and made available for research. It is more precise than play-by-play derived approximations, but is not fully transparent (the exact model specifications are not public). Where this distinction matters for interpretation, the individual metric notes say so.
Metrics sourced from nflverse / nflfastR are open-source and reproducible. Metrics sourced from Next Gen Stats (NGS) use the NFL's proprietary player-tracking infrastructure — optical tracking of every player on every play. NGS data is licensed to the NFL and made available for research. It is more precise than play-by-play derived approximations, but is not fully transparent (the exact model specifications are not public). Where this distinction matters for interpretation, the individual metric notes say so.