Methodology

How CFBTrack turns raw college football data into readable rankings and tools.

CFBTrack pages are designed to stay explainable. This page outlines how the site interprets team, player, and season data so the numbers remain auditable instead of feeling opaque.

View data sourcesBrowse analytics pages

What this page covers

  • Ranking models

    Best-teams, comparison, and leaderboard pages blend raw production with context, not one isolated stat.

  • Interpretation rules

    Lower-is-better and higher-is-better metrics are labeled on-page and normalized before they are blended.

  • Fallback behavior

    When a dataset is incomplete, pages should say so directly instead of inventing certainty.

How weighted pages are built

Model-driven pages such as Best Teams by Program combine multiple season-level components so one great trait does not automatically dominate the result. Offensive quality, defensive quality, schedule-aware profile strength, margin, postseason context, and broader quality signals are normalized and then weighted into one score.

  • Offense and defense ranks are rank-style measures where smaller numbers are better.
  • Schedule/profile rank blends schedule-aware SRS and SP-style overall strength, where #1 means the strongest profile in the compared pool.
  • Weighted pages expose slider controls when a page is designed for debate rather than one fixed answer.

How comparison and scatter tools work

Comparison pages use compact season-level view models so the charts stay readable and the server payload stays small. Scatter pages pair two metrics at a time, compute a simple trendline, and then surface text summaries and notable outliers so the page remains understandable even without chart interaction.

  • If the newest season has incomplete data, the tool should fall back to the most recent season with usable coverage.
  • Empty states should explain what filters removed the field and offer a clear reset path.
  • Known kickoff times and explicit metric labels are preferred over ambiguous defaults.

How player profile metrics work

Player profile pages choose a position-aware primary metric for chart defaults, split summaries, and profile cards. Quarterbacks use total offense, running backs lean on scrimmage yards, receivers and tight ends use receiving production, defenders use disruption or tackling signals, and fallback cards use touchdowns or the strongest available volume stat.

  • For quarterback profile cards, total offense means passing yards plus rushing yards; receiving and return yards stay in their own stat families.
  • Stats search rows are scoped to a player, team, season, and season type so transfer stops remain auditable; player season leaderboards can aggregate transfer seasons by player while retaining team labels.
  • Efficiency, usage, takeover, and season-value cards are CFBTrack-derived display scores from available source stat rows. They are not NCAA official ranking categories and should be hidden or caveated when supporting rows are missing.

How to read the outputs

Most CFBTrack pages try to keep the visible explanation close to the result. If a table, card, or chart highlights a team, the page should also expose enough adjacent context to understand why it landed there and how to verify it.

  • Use linked team, season, and stat pages to trace a summary back to the underlying record.
  • Treat future-season or partial-season records as provisional until the source data settles.
  • When a page says a metric is a rank, assume a deterministic sorted order with a stable tie-breaker.
  • National title pages count title seasons as one champion per team-year; selector labels stay visible, and conference breakdowns use historical affiliation where available.
  • Representation counts use known, matched values; unresolved placeholder rows can remain in raw totals without inflating unique school, state, conference, or program counts.

Route families covered by this methodology

The same reading rules show up across several route families. Leaderboards answer who led a selected sample, comparison tools answer how two or more teams stack up, history pages answer durable record questions, and map or media pages answer geography and distribution questions. Each family needs a slightly different level of caution.

  • Leaderboard pages should be read with season, week, conference, and season-type filters in mind.
  • Comparison and scatter pages are strongest when the compared teams share enough context to make the result meaningful.
  • History and records pages are more stable, but era rules, title selectors, all-time record conventions, and historical conference structures still matter.
  • Map, media, video, and recruiting pages can depend on specialty datasets that may not refresh on the same cadence as core game data.

What CFBTrack avoids claiming

The site should not turn a single number into a complete verdict. A high ranking, strong chart position, or unusual outlier is an invitation to inspect the page context, not a guarantee that one team, player, coach, or program is better in every reasonable frame.

  • Predictive-sounding results should be described as model outputs, not certainties.
  • Partial-season and future-season records should stay provisional until source coverage settles.
  • Missing source fields should create visible caveats or hidden modules, not invented values.

Deep dives

Methodology deep dives by model and tool

These pages explain the specific models and route behaviors users are most likely to question from an analytics page.

Dominance ScoreLearn how CFBTrack blends efficiency, schedule-aware strength, margin, winning, postseason, and advanced-quality signals into a dominance score.Weather ImpactLearn how CFBTrack groups weather conditions and compares scoring, style, and upset profiles across college football games.Talent UsageLearn how CFBTrack compares recruiting input, on-field results, and NFL draft output to estimate talent usage.Team ComparisonLearn how CFBTrack compares teams with compact season models, metric labels, trendlines, and outlier context.Best SeasonsLearn how CFBTrack chooses each program's best modern season and flags narrow, debatable gaps.Incomplete SeasonsLearn how CFBTrack treats provisional seasons, missing fields, fallback years, and small samples.