Team Strength Engine
An explainable, provisional model that normalizes team performance per game and shrinks small samples toward neutral. It is a research ranking, not a calibrated win probability.
How it works
Raw index = point differential per game + 0.025 × yard differential per game + 2 × turnover margin per game
Exploratory strength = raw index × games / (games + 6)
The coefficients and shrinkage are initial engineering assumptions. They must be backtested and calibrated on historical seasons before any predictive claim or pick can be made.
Required inputs
| Field | Meaning |
|---|---|
| games | Completed games in the sample |
| points_for / points_against | Total points scored and conceded |
| yards_for / yards_against | Total yards gained and allowed |
| turnover_margin | Takeaways minus giveaways across sample |
Development status
The administrator-only REST endpoint POST /wp-json/patternpicks/v1/football/team-model has been saved as a WPCode snippet and requires administrator activation. It calculates from explicitly supplied team totals; automatic NFL ingestion, opponent adjustments, injury context, odds integration and backtesting are not yet connected.
No scores, confidence percentages or recommended bets will be displayed without verified inputs and a validated model.