The NFL Usage Stats Glossary: Every Metric, Plainly Defined

Every metric below is defined in plain language, with a note on what it is good for and what it will mislead you about. The organizing principle is simple: usage metrics measure opportunity and are stable, efficiency metrics measure outcomes and are noisy. When the two disagree over a short sample, the usage number is almost always the one telling you about next week.

The receiving stats in particular are stages of a single funnel — snaps, then routes, then targets, then catches — and each stage filters the one before it. Reading them in order is the difference between analysis and guessing.

Usage and opportunity

These describe how much of an offense a player is getting. They are the foundation, they are the most predictive, and most of them are free.

• Snap count — The raw number of offensive plays a player was on the field for in a game. Good for: establishing that a player has a role at all. Not good for: comparing between games, since offenses run wildly different numbers of plays.
• Snap share — Snap count divided by the team’s total offensive snaps, as a percentage. The first number to check on any player at any position. Good for: catching role changes before the box score does. Not good for: distinguishing a snap at the goal line from a snap in garbage time — see what snap share does and does not tell you.
• Dropbacks — Plays on which the offense intended to pass, including sacks and scrambles as well as attempts. The correct denominator for anything about the passing game. Good for: sizing the passing-game pie. Not good for: nothing much — it is a raw count and behaves like one.
• Route participation — Routes run divided by team dropbacks. Distinguishes a pass-catcher from a player who is merely on the field. Good for: exposing tight ends whose snap counts are mostly blocking, covered in full in route participation. Not good for: free analysis — it comes from charting, not the official record.
• Target — A pass thrown in a receiver’s direction, whether caught, dropped, defended or overthrown. Good for: the currency of receiving production; nothing happens without one. Not good for: judging accuracy or intent — an uncatchable throw counts the same as a perfect one.
• Target share — A player’s targets divided by his team’s pass attempts. Roughly 25% or more is elite; under 15% produces weeks you cannot start. Good for: ranking pass-catchers within an offense, as target share and air yards sets out. Not good for: comparing across teams without adjusting for how often each one throws.
• Air yards — The distance a target travels past the line of scrimmage, credited whether or not the pass is completed. Good for: measuring offensive intent, unfiltered by outcome. Not good for: measuring production — air yards on incompletions are worth nothing.
• aDOT — Average depth of target: air yards divided by targets. Under 6 is a screen-and-checkdown profile; 14 and up is a field-stretcher. Good for: separating floor-heavy possession receivers from boom-or-bust deep threats. Not good for: ranking players — neither end of the range is inherently better.
• WOPR — Weighted opportunity rating, created by analyst Josh Hermsmeyer. The formula is 1.5 × target share + 0.7 × air yards share, blending volume and depth into one number. Good for: comparing receivers with different target profiles on one scale. Not good for: precision — small gaps between players are noise.
• TPRR — Targets per route run: targets divided by routes run. Answers how often the quarterback looks a player’s way when he is actually in the pattern. Good for: finding target earners stuck in small roles. Not good for: projecting output on its own, since a great rate on 12 routes is still 12 routes.
• Carry share — A running back’s carries divided by his team’s total rushing attempts. Good for: mapping a backfield hierarchy quickly. Not good for: valuing passing-down backs, whose worth is mostly in the receiving game.
• Opportunity share — A back’s carries plus targets divided by the team’s combined carries and running back targets. The single best summary of a backfield role. Good for: comparing backs in different offenses. Not good for: distinguishing which touches are valuable — a carry at the 1 and a carry at the 20 count identically.
• Touches — Carries plus receptions. The old-fashioned volume measure and still a decent one. Good for: a quick sanity check on workload. Not good for: anything requiring nuance; touches are not fungible.
• High-value touches — The subset of touches worth disproportionate fantasy points; the common definition is receptions plus carries inside the 10-yard line, though analysts draw the line differently. Good for: explaining why a back with fewer touches outscores one with more. Not good for: cross-source comparison, given the varying definitions.
• Red-zone share — A player’s carries and targets inside the opponent’s 20 divided by his team’s total. Track the 10 and the 5 separately where you can. Good for: predicting touchdowns, as red-zone usage explains at length. Not good for: stable reads on small samples — a team may get only four or five red-zone trips a game.
• Vacated targets (target vacuum) — The targets, carries or routes left behind when a player leaves via injury, trade or free agency, and which someone on the roster will absorb. Good for: identifying who is about to be promoted, especially when reading preseason snap counts. Not good for: assuming a clean transfer — vacated work often disperses across several players or disappears into a scheme change.

Efficiency and production

These describe what happened with the opportunities. They are more interesting, more quoted, and much less predictive over the samples you actually have.

• Yards per route run (YPRR) — Receiving yards divided by routes run. The standard way to compare receivers with very different playing time. Good for: finding productive players in small roles. Not good for: short samples, where one long catch swings the number badly.
• YAC — Yards after catch: yardage gained past the reception point. Good for: understanding how a receiver generates production and whether his aDOT understates his value. Not good for: prediction — YAC is heavily influenced by scheme, blocking and missed tackles.
• EPA — Expected points added. Every game state (down, distance, field position, time) carries an average expected point value for the offense based on historical outcomes; EPA is the change in that value caused by a single play. Good for: grading plays and teams in a way that respects situation. Not good for: isolating an individual skill player, since EPA is credited to a whole play, not distributed among the eleven people who ran it.
• Success rate — The share of plays that count as successful. Definitions vary: one common version calls a play successful if it gains a set fraction of the yards needed — roughly 40–45% on first down, 60% on second, all of it on third or fourth — while the EPA-based version simply counts plays with positive EPA. Good for: measuring consistency rather than explosiveness. Not good for: cross-source comparison, so check which definition you are reading.
• DVOA — Defense-adjusted value over average. A rate statistic that grades each play against a league-average baseline for the same situation and adjusts for opponent quality. Originated at Football Outsiders and now published by FTN. Good for: comparing team and unit strength with schedule effects stripped out. Not good for: week-to-week fantasy decisions about individual players.
• Expected fantasy points — A usage-weighted estimate of what a player’s opportunities should have been worth, built by assigning an average point value to each carry and target based on situation and field position. Good for: separating a good process from a good result. Not good for: treating as precise, since every provider builds the model differently.
• Expected touchdowns — The same idea narrowed to scoring: each opportunity gets a touchdown probability based on where and how it happened, and the sum is what the usage supported. Good for: identifying regression candidates in both directions. Not good for: fine distinctions between players whose numbers are close.

Situation and game context

These set the size of the pie. They apply to whole offenses and therefore to every player on them at once.

• Play-action rate — The share of dropbacks that include a run fake. Good for: context on a quarterback’s efficiency and on the route trees available to tight ends. Not good for: projecting individual usage directly.
• Pace — How quickly an offense runs plays, usually expressed as seconds per play or total plays per game. Good for: estimating how many opportunities an offense will generate. Not good for: raw comparison, because trailing teams play fast by necessity.
• Neutral-game-script pace — Pace measured only on plays where the score is close, filtering out hurry-up and clock-killing. Good for: the honest read on how fast a team actually wants to play. Not good for: projecting a specific game where you expect a lopsided result.
• PROE — Pass rate over expectation: a team’s actual pass rate minus the rate a model expects given down, distance, field position, score and time remaining. Positive means a team throws more than the situation calls for. Good for: identifying offenses that will support pass-catchers regardless of game flow. Not good for: week-level prediction, since one script-driven game can swing it.
• Game script — The flow of a game as dictated by the score: leading teams run and drain clock, trailing teams throw. Good for: explaining after the fact why usage looked strange. Not good for: forecasting with confidence, which is why usage trends beat single-game readings.
• Garbage time — Plays late in a decided game, when defenses concede yardage and personnel is not what it would be in a competitive situation. Good for: knowing which production to discount. Not good for: dismissing outright — garbage-time points still count in your league.

Roles and depth-chart language

• Bell-cow — A running back who handles the overwhelming majority of his team’s backfield work on all three downs, typically 70% or more of snaps. Good for: a weekly floor almost independent of game script. Not good for: assuming permanence — bell-cow roles collapse quickly with injury or a hot backup.
• Committee — A backfield split between two or more backs, often by situation rather than evenly. Good for: nothing, from a fantasy manager’s point of view. Not good for: projecting weekly output, which is why backfield committees need their own reading method.
• Handcuff — The backup who would inherit a starter’s workload if the starter were unavailable. Good for: insurance on a roster built around one back, and a genuine asset in deep leagues. Not good for: roster spots in shallow leagues, where the opportunity cost is too high.

Betting terms

• Player prop line — A market on an individual player’s statistical output — receiving yards, receptions, rushing attempts — usually offered as an over/under with a price attached to each side. Good for: expressing a usage view on a single player. Not good for: casual play, because the margin built into most player-prop pricing is wider than on main markets.
• Anytime touchdown prop — A market on whether a player scores at least one touchdown. The betting expression of red-zone usage analysis. Good for: translating a goal-line role into a position. Not good for: the assumption that identifying a likely scorer is the same as finding value, since the price already reflects the obvious candidates.
• Implied probability — The break-even win rate a price represents. For decimal odds it is 1 ÷ decimal, so 1.91 implies 1 ÷ 1.91 ≈ 52.4%. For negative American odds it is |odds| ÷ (|odds| + 100), so −115 implies 115 ÷ 215 ≈ 53.5%. Good for: knowing exactly how often you must be right to break even. Not good for: reading as a true probability, since the sum of implied probabilities across a market exceeds 100% by the operator’s margin.

How to use these together

The order is the point. Start with snap share to confirm a player is on the field. Move to route participation or carry share to confirm what he does when he is. Then target share or opportunity share, then red-zone usage for scoring, and only then look at efficiency to sanity-check whether a result was earned. Running that sequence in the same order every week is the whole of the usage-based fantasy process, and it is what makes the difference between spotting a role change on Tuesday and reading about it on Thursday.

The one habit worth adding to any of these definitions: write down how many games your number covers. Most disagreements about players are actually disagreements about sample size, and a four-week share and a full-season share are different claims that deserve different amounts of confidence. That distinction matters most in-season, when waiver decisions built on usage signals are made on three or four games of evidence and have to be made anyway.