# Metrics and definitions This page is the canonical glossary for terms used across the CFBD API. Model pages add the context specific to their outputs, while this page defines the shared building blocks. ## Reading API values - **Rates and probabilities** are generally returned as decimals. A value of `0.64` represents 64%, not 0.64%. - **Average** values are normalized per included play or opportunity. **Total** or **cumulative** values are summed across the included sample. - **Offense** fields describe value created by the team in possession. **Defense** or **allowed** fields describe what opponents produced against that defense. Lower allowed values are generally better. - A **rating** is a model value. A **ranking** is the ordered position produced by sorting those values. - `null` means the value is unavailable or not applicable. It should not be converted to zero without an explicit analytical reason. ## Expected points and play value ### Predicted Points (PP) CFBD's estimate of the expected value of the next score from a given down, distance, and field position. Positive values favor the offense; negative values indicate that the opponent is more likely to score next. ### Predicted Points Added (PPA) The change in Predicted Points attributable to a play, from the offense's perspective. Positive PPA means the play improved the offense's expected scoring position. Negative PPA means it lost value. PPA is CFBD's implementation of the broader Expected Points Added (EPA) concept. ### Explosiveness Average PPA on successful plays. This captures the magnitude of a team's successful plays rather than how often those plays occur. See [PPA methodology](./ppa) for calculation context and interpretation. ## Situational definitions ### Success rate The share of plays classified as successful. A successful play gains at least 50% of the yards needed on first down, 70% on second down, and 100% on third or fourth down. Scoring plays are successful unless the recorded play type explicitly represents a failed attempt. ### Standard and passing downs Stored historical and advanced metrics classify second-and-8 or longer and third- or fourth-and-5 or longer as passing downs. All other situations are standard downs. Some live values are calculated independently from the stored historical pipeline. When a live response supplies its own down classification, use the returned value rather than recomputing the split. ### Garbage time Garbage time identifies plays where the pre-play margin is greater than 38 points in the second quarter, 28 points in the third quarter, or 22 points in the fourth quarter. A game can leave garbage time if the margin narrows below the applicable threshold. ### Scoring opportunity A drive in which the offense reaches the opponent's 40-yard line or closer. Points per opportunity is the average number of points scored on those drives. ## Rushing-line metrics ### Line yards A weighted rushing-yard measure intended to estimate the offensive line's share of a run. Losses count at 120%, yards 0 through 4 at 100%, yards 5 through 10 at 50%, and yards beyond 10 do not add line yards. ### Second-level yards Rushing yards gained between 5 and 10 yards beyond the line of scrimmage. ### Open-field yards Rushing yards gained more than 10 yards beyond the line of scrimmage. ### Highlight yards The portion of rushing production credited after the line-yards component, intended to describe what the ball carrier creates in the second level and open field. ### Stuff rate The share of rushing attempts stopped at or behind the line of scrimmage. ### Power success The share of short-yardage runs needing two or fewer yards that meet the success definition. ## Disruption and involvement ### Havoc rate The share of defensive plays that produce a tackle for loss, forced fumble, interception, or pass breakup. **Front-seven havoc** uses tackles for loss and forced fumbles. **Defensive back havoc** uses passes defended and interceptions. ### Usage rate The share of a team's relevant plays involving a player. Overall, passing, rushing, standard-down, and passing-down usage use the corresponding team play total as the denominator. ## Win probability terms ### Pregame win probability An estimate of the home team's chance of winning before kickoff, derived from the available point spread. ### In-game win probability An estimate of the home team's chance of winning at a specific game state. The API returns a decimal from 0 to 1. ### Postgame win probability An estimate of how often a team would be expected to win with similar underlying game performance. It is separate from both the final result and the play-by-play in-game model. ### Excitement Index A summary of win-probability movement during a game. Larger and more frequent swings produce a higher value. Because it depends on the underlying win probabilities, model changes can affect comparisons across eras. See [Win probability methodology](./win-probability) for the three distinct models and their coverage.