BetWise Match Lab compares a current college football game's full pregame statistical fingerprint against historically similar CFB matchups — then shows what actually happened. Free College Football Historical Matchup Research.
Select a current college football game to view its pregame fingerprint. Live games are read through The Odds API on the server. When the integration is not configured, a setup state is shown — no games are fabricated.
Choose one market: Spread, Moneyline, or Over/Under. Each market uses its own configurable similarity model. Props are not supported in Phase 1.
Side-by-side pregame profile across roughly 50 features in seven groups: Team Strength, Offensive Profile, Defensive Profile, Line/Possession Profile, Situational Profile, Market Profile, and Personnel Profile. Only reliable, pregame-available fields are shown. Unavailable values display N/A — never a fabricated zero.
A composite statistical similarity score (0–100) against historically analogous pregame conditions. This is a statistical similarity measure, not a win probability. The score appears once a historical CFB dataset is ingested and the market model is calibrated.
Up to five historical CFB games whose pregame fingerprints most resemble today's matchup. No results are forced — if no analog qualifies, a no-match state is shown. Each card supports matchup, date, season/week, venue, score, historical spread/total, key pregame metrics, reasons matched, biggest differences, final score, winner, and ATS and O/U outcomes — populated only from real data.
Aggregate counts, records, margins, scores and similarity bands across the ranked analog set. Small samples are never expressed as precise probabilities.
Current Game → Pregame Fingerprint → Normalize Features → Market-Specific Similarity Model → Search Historical Pregame Database → Rank Closest Analogs → Display Historical Outcomes.
Pregame snapshots and postgame results are strictly separated. Historical matching reads only information available before each game's kickoff, preventing look-ahead and data leakage. The engine supports weighted standardized distance, cosine similarity, Mahalanobis-style comparison and nearest-neighbor retrieval, with a model/version layer for feature weights, included features, market weighting, formula and thresholds. Separate configurable models exist for Spread, Moneyline and Total. Weights are not finalized in Phase 1.
Server-side integration points for CFBD_API_KEY (CollegeFootballData) and ODDS_API_KEY (The Odds API) keep credentials off the client. The historical CFB analog database has not been ingested yet; similarity search returns an explicit awaiting-ingestion state until then.
Historical similarity does not guarantee the same future result. Match Lab compares historical pregame conditions and outcomes for research purposes. Small historical samples should not be interpreted as precise probabilities. For research and educational purposes only. Not financial advice or a guarantee of any outcome. Must be of legal age where applicable. If you or someone you know has a gambling problem, call 1-800-GAMBLER.
Research Guides · CFB Research Guide · Why BetWise? · Odds Converter