Glossary

The NCAA's two Division I football tiers. FBS (Football Bowl Subdivision) is the top level — the ~136 programs eligible for the College Football Playoff and major bowls; FCS (Football Championship Subdivision) plays its own smaller-scholarship tier with its own 24-team playoff. This site stores it as teams.level_code (values FBS/FCS/D2/D3) and most site-wide rankings, power ratings, and model outputs are FBS-only.

Shorthand for the two informal tiers inside FBS. P4 ("Power Four") = the ACC, Big Ten, Big 12, and SEC (plus independent Notre Dame, which this site treats as P4) — the four conferences with the deepest revenue and recruiting bases. G5 ("Group of Five") = every other FBS conference (AAC, Conference USA, MAC, Mountain West, Sun Belt). Older coverage sometimes says "P5"/"Power Five" — that was accurate before the Pac-12's 2024 collapse folded it down to four; this site's current tier logic (e.g. claim_miner/templates.py) uses P4.

The College Football Playoff — the sport's postseason championship field, expanded to 12 teams starting with the 2024 season. This site's playoff-forecast tooling projects the field and each team's bid odds well before Selection Sunday, tracking movement week to week rather than just a final snapshot.

An external advanced-stats power rating (Bill Connelly / ESPN) that this site displays alongside its own CFB Zeitgeist power rating, FPI, Elo, and SRS in the team-page "Model Consensus" module — a side-by-side read on where the national models agree or disagree with this site's own rank. SP+ itself is not computed by this site; it's ingested and shown for comparison only.

Sportsbooks build a profit margin ("the vig" or "juice") into their posted odds, so raw implied probabilities from both sides of a bet always sum to more than 100%. De-vigging removes that margin — converting American odds to implied probabilities, then rescaling the whole field so it sums to exactly 100% — to get a fair market read. This site's market.py does this (devig_field()) for futures markets like CFP championship odds.

The uncertainty parameter this site's power model uses to convert a predicted point spread into a win probability (p_home = Φ(spread / σ) — a larger σ means a given spread implies a less confident probability). As of the current build it's population-split rather than one flat number: σ = 12.4 when both teams are FBS (the walk-forward Brier-optimal value for FBS-vs-FBS games), σ = 10.0 for any game touching a sub-FBS team (FCS/lower-level blowouts pull the all-games optimum lower).

The site's flagship 0-100 player-season score — one number per player per season describing how good that season was, position-aware, meant to play the same role ESPN's Total QBR plays for quarterbacks but for every offensive skill position. It's built from real-football-unit value (a blend of opponent-adjusted EPA and box-score production, shrunk toward a position mean for small samples) mapped onto a fixed scale where ~50 is a median qualified starter, ~75 is All-Conference caliber, and ~90 is All-America caliber. It is not a career-value ladder or a draft projection, and it is a different number from the team-level "247 Talent Composite" or a player's individual "247 composite" recruiting rating — see those entries below.

Every named metric on this site that depends on sample size or data depth ships with a confidence band so the reader knows how much to trust the number, instead of a bare figure with no context. High (blue, ✓) = data comfortably above the calibrated threshold for that metric's domain; medium (gold, •) = normal range; low (gray, ?, shown italic) = thin data — the number is real but preliminary. Below the lowest threshold, the metric is suppressed rather than shown unreliably. Thresholds are recalibrated from the live data distribution, not fixed arbitrarily.

This site's inline honesty tags for AI-assisted editorial copy: a blue ✓ fact tag marks a claim anchored directly to verifiable data (a stat, a result, a recorded quote); a gold ~ our read tag marks the site's own interpretation or synthesis layered on top of that data. The two are styled to never look alike on purpose — the color and glyph both change, not just the wording — so a reader can tell at a glance which parts of a card are sourced and which are editorial judgment.

Recruiting shorthand for a 4-star or 5-star high school prospect (the top two tiers of the standard 5-star scale) — a "blue-chip prospect" or "blue-chip recruit." A program's "blue-chip ratio" (share of roster that arrived as a 4- or 5-star) is a commonly-cited proxy for a program's talent ceiling.

The recruiting industry's aggregated star rating for an individual player — 247Sports' composite of multiple outlets' evaluations, expressed on a roughly 0-to-1 scale (higher is better). This site ingests it via CFBD for both high-school recruits and transfer-portal entrants, and for portal scoring purposes converts it to a 1-to-5 star-equivalent talent score (5-star ≥ 0.98, 4-star 0.90-0.97, 3-star 0.80-0.89, below that 2-star). This is a per-player number — not to be confused with the team-level "Talent Composite" below.

A team-level number, distinct from an individual player's "247 composite" rating above: it's the aggregate 247Sports/On3-style talent ranking of a program's entire roster, shown on team pages as a score plus a national rank. Where "247 composite" grades one recruit or transfer, Talent Composite grades the whole roster's cumulative recruiting investment — and both are different again from the performance-based "CFB Index score," which grades how a player or team actually played, not how they were recruited.

The Transfer Portal Heat Index's ranking metric: talent-weighted portal arrivals minus talent-weighted portal departures for a program over a rolling window. Each transfer's weight comes from their 247 composite rating scaled by a position-value multiplier (quarterbacks and cover corners count for more than running backs, reflecting real positional scarcity research), then decayed by depth — a program's second, third, and Nth transfer at the same position count for progressively less, since a roster can only start one. A positive net Δ means a program's incoming talent outweighs what it lost; the Portal Heat board ranks the whole FBS by this number.