The GridironEdge model predicts college football spreads and totals using a multi-layer algorithm that blends power ratings, drive efficiency, opponent adjustments, tempo, red-zone metrics, and market comparison.
For every FBS team, we maintain a neutral-field Power Rating built from:
Early-season priors decay over time, so current form matters more as the schedule progresses. Extreme outlier games are winsorized instead of letting one crazy box score wreck the rating.
Once each team has a neutral-field rating, we apply matchup-specific adjustments:
This produces our Model Spread and Model Total for a given matchup on a specific day.
The core of GridironEdge is comparing our projection to the market spread. We always express the spread from the favorite's perspective (Team 1). Edge is computed as:
Edge_T1 = Vegas_Spread_T1 − Model_Spread_T1
A large positive or negative value signals that the market and model strongly disagree. Sign handling is validated against real games so that a “flip” in the pick matches what actually happens on the field instead of being a formula mistake.
Not every edge becomes a Top Pick. We filter based on:
Totals are only considered when both teams show explosive offense and consistently poor defense, and even then we lean to overs. We avoid unders due to volatility and small mistakes skewing totals.
Every Top Pick is timestamped and logged. At a high level:
As the platform matures, users will be able to download CSVs of all historical picks, edges, and results for independent verification.