Glossary

Share of plays that stay on schedule.

≥50% of yds-to-go on 1st · ≥70% on 2nd
· 100% on 3rd/4th

How big the successful plays are. Separates teams that grind from teams that hit chunks.

mean EPA on successful plays

Passing downs are when the defense expects a pass. Everything else is a standard down.

PD = 2nd & 8+ · 3rd/4th & 5+

Points per drive that reaches the opponent’s 40. Measures finishing, not just moving the ball.

pts ÷ drives reaching opp 40

Share of drives reaching the opponent 20 that end in a touchdown. Offense scores it, defense allows it.

TD drives ÷ drives reaching opp 20

Share of plays where the defense makes a disruptive play.

(TFL + SACK + PBU + INT + FF) ÷ plays

Rushing yards credited to the offensive line. Losses are weighted heavier and long gains are capped.

−1.2×(loss) + 1.0×(0–4) + 0.5×(5–10) + 0×(11+)

Plays run after the outcome is effectively decided. They are excluded by default.

Score margin > 43 (Q1), 37 (Q2), 27 (Q3), 21 (Q4)

Expected points added per game by the kicking game: field goals, punts, kickoffs and returns, each from the team’s own side. Never opponent-adjusted.

(FG + PUNT + KO + RET EPA) ÷ games

Points per game from turnover margin beyond what fumbles and passes defensed predict. Big numbers either way tend to regress.

(TO margin − xTO margin) × pts/TO ÷ games

Completion % over expected, given target depth, down, distance and field position.

comp% − xComp%

Average depth of target, and the share of attempts thrown 20+ air yards. Parsed from play-by-play on every dropback.

air yds ÷ targets · att 20+ air yds ÷ att

Passes defensed against a QB (INTs + breakups) per attempt, and the interceptions those passes would produce on average.

(INT + PBU) ÷ att · xINT = 0.188 × (INT + PBU)

Pass rate over expected. How often a team throws compared with an average team in the same situation.

pass% − xPass% · neutral WP 20–80%

Game clock elapsed between snaps on the same drive, across all situations.

Σ clock between snaps ÷ snaps

Every play is adjusted for the strength of the opposing unit, solved by ridge regression across all FBS games.

EPA ~ off + def + hfa · λ = 125

Gap between our projected spread and the consensus market line. Edges of 2+ points are highlighted.

|model spread − market spread|

Closing line value, reverse line movement (the line moves against the ticket majority) and steam (4+ books in 5 min).

CLV = bet − close · STEAM = 4+ books in 5 min

How often a team wins if the same game is replayed from its own plays. Tells a dominant win from a lucky one.

P(win) over 10,000 resamples of each side’s EPA plays

Where a team stands for the CFP. Current status is a rule of thumb from losses; projected status buckets the season sims’ playoff odds.

lock ≥ 95% · likely ≥ 70% · bubble ≥ 30% · long shot > 0