GE

GRIDIRON EDGEHISTORICAL MONEYLINE LAB

← Back to Gridiron Edge

NCAA FOOTBALL · BACKTESTING

Find the edge. Verify the price.

Compare outright wins, favorite spread covers, real historical moneyline returns, and your saved model forecasts without substituting fictional odds.

Historical analysis · not live bets

1. Load completed seasons

CFBD historical /lines · server-side key

From seasonThrough seasonSportsbook

Connect your encrypted CFBD_API_KEY in Cloudflare to load actual lines. You can also import an odds CSV below.

Import historical odds CSV or normalized JSONNo local odds file chosenImport Gridiron Edge model snapshots CSVNo model snapshots imported

Odds data requires an actual quoted moneyline. CFBD's available lines are generally historical closing records, but quote timestamps are not independently certified. Game rows without odds are excluded from moneyline ROI rather than assigned -200.

2. Historical model replay

CFBD game-level advanced stats · no market price in predictions

New: reconstruct 2023–2025 pregame inputs automatically. Replay uses only prior finished games with an 8-hour kickoff buffer. Its formula is an experimental proxy based on Gridiron Edge's older matchup philosophy (rush 30%, trenches 20%, pass 20%, situation 15%, context 15%). It is not a reproduction of your current v1.7 engine. Historical features and prices have not been timestamp-certified.

Minimum earlier games per team

Earlier blowoutsExclude margins over 28

Load historical odds above, then generate season-by-season forecasts. CFBD advanced game stats may require sufficient plan access and API quota.

No replay generated. Baseline moneyline tests can still run without model data.

A reconstructed forecast has a historical data cutoff, but it was generated today. We deliberately leave its generatedAt blank; ticking the timestamp-certified filter below will exclude it. To claim a true out-of-sample result, freeze this formula and validate later untouched games.

3. Choose backtest filters

Same game set, different strategies

Favorite spread fromToStake per bet ($)Custom probability ≥EV edge ≥ (points)

Require a timestamped model snapshot before kickoff (unticked: archived backtests remain exploratory)

4. Results by strategy

No odds loaded

Games loaded—

Eligible priced favorites—

With model predictions—

Timestamp-checked snapshots—

Strategy	Bets	W–L	Hit rate	Moneyline profit	Moneyline ROI	ATS profit (-110)	Max ML drawdown
Load CFBD historical prices or import an odds file to begin.

Profit is calculated per actual same-provider favorite moneyline, with the stake you set above. ATS assumes a hypothetical -110 price and ignores pushes. Historical model files are not independent, pregame-verified forecasts unless their prediction timestamps precede kickoff.

Season-by-season

Season	Bets	Wins	ML profit	ATS profit
—

Data quality and interpretation

No data loaded. Baseline tests don't require model predictions; model strategies do.

A historical simulation is not proof of future profitability. Seasons used to tune Gridiron Edge are in-sample. For out-of-sample validation, freeze the formula, generate forecasts before kickoff, and test on later games.

Individual games

Select a strategy to view qualifying picks.

Date / week	Matchup	Favorite	Spread	ML	Model P	Edge	Result	ML profit
—