Methodology
Last updated: August 2026
PickGauge doesn't run its own prediction model from scratch. Model # is a weighted average of computer prediction systems and market data pulled from outside sources, plus optional context from a sports-data provider. This page explains what those sources are, what they're used for, and — for the sources this matters most for — that PickGauge isn't affiliated with or endorsed by any of them.
What Model # actually is
Every prediction system you enable in the Prediction Systems panel becomes one number in a weighted average (default: equal weight, adjustable per system). That average — Model # — is compared against the market line to produce Edge and Cover %. It's a simple, transparent average, not a trained model of PickGauge's own, and it's only ever as good as the systems feeding into it.
Data sources
The great majority of Model #'s inputs — every named computer rating and prediction system in the checklist (Sagarin variants, Massey Ratings, ESPN FPI, and the rest) — come from thepredictiontracker.com, which aggregates weekly predictions from dozens of independent computer rating systems. PickGauge is not affiliated with or endorsed by thepredictiontracker.com. If you run one of the systems it tracks and have questions about how your data is used here, please reach out.
Live and historical sportsbook spreads shown on the Edge Board come from The Odds API, a commercial odds-data provider PickGauge has a paid account with.
Live scores, power ratings (CORE/SP+/FPI/Elo/SRS), Matchup Intelligence, postgame box scores, and historical betting-line checks all come from CollegeFootballData.com, via a paid CFBD API account. All of this is shown as context around a game — it never changes Model #, Edge, Cover %, or EV. The one deliberate exception: SP+ and CORE can optionally be enabled as real, weighted inputs to Model # (derived from the underlying CFBD rating plus a fixed home-field constant), exactly like any thepredictiontracker.com system — every other CFBD feature stays context-only.
What PickGauge itself computes
A few things shown throughout the app are PickGauge's own math, not pulled from an outside source:
- Edge, Cover %, and EV — derived directly from Model # vs. the market line, using a cover-probability curve fitted from thousands of real historical FBS games.
- CLV (closing-line value) — the market's move between your pick and the retained pre-kick closing line.
- The ★ Top 10 backtest ranking — a real 2-year backtest (40% ATS% / 30% MAE / 30% |Bias|) of every enabled prediction system's own historical accuracy, used only to flag which systems have actually performed well — never to auto-weight Model # itself.