Backtest Explanation
The NFL results are built to answer the customer question that matters: if this side and totals portfolio had been operating in the tested seasons, and a member risked $100 on every qualified play at the realistic decision-time price, what would have happened?
Walk-Forward Method
A walk-forward backtest moves through the season in order. The model is not allowed to know the final standings, final third-order rankings, closing score, or any information that happened after kickoff. Each game is treated like a live decision point: use the information available at the time, compare the model number to the betting market, apply the Three Check rules, apply the portfolio governor, and then grade the result after the game is played.
The current NFL side and totals work uses realistic decision pricing. The system builds from the early market baseline and grades qualified plays from the best available sportsbook price near the decision window, roughly four hours before kickoff when that snapshot is available. That is closer to how customers would actually bet than using a stale opening number.
The results page is chronological on purpose. Each tested season is shown by itself first, then the bet types are shown inside that season. This avoids hiding a weaker or stronger year inside a multi-season total and lets customers see how the process performed year by year.
What The Model Compared
The model's job is to create a fair number. For spreads, that means estimating what the spread should be. For totals, that means estimating what the game or first-half total should be. A sportsbook line only becomes interesting if the model shows enough value against the available market price.
Three Check Process
The Three Check process is the discipline layer. It keeps the card from becoming every model opinion and forces the play to survive value, market, and football-context checks before it reaches the customer record.
The model must show a meaningful edge compared with the sportsbook spread, first-half spread, game total, or first-half total.
The play is checked against market behavior so the card is not built only on isolated model disagreement.
The matchup still has to make football sense through the governor rules, so fragile or dangerous edges can be filtered out.
Portfolio Governor
The newest NFL product improvement is the overlap governor. If a first-half model and a full-game model both fire on the same game, the card does not blindly bet both. The historical overlap test decides which lane should get priority.
When tightened full-game ATS and premium first-half ATS both fired, first-half priority produced the best portfolio efficiency. The official side portfolio therefore keeps first-half ATS in same-game overlap spots.
When full-game totals and first-half totals both fired, full-game priority produced the best portfolio efficiency. The official totals portfolio therefore keeps the full-game total in same-game overlap spots.
The results page still shows the underlying full-game and first-half lanes for transparency, but the season combined card uses only the governed side portfolio plus the governed totals portfolio. The underlying lanes are not double-counted.
CLV And Pricing
Closing-line value, or CLV, measures whether the bet beat the later market number. Positive CLV is useful because it tells us the model was generally ahead of market movement, even before knowing the final score.
The NFL sides card now has true-close spread CLV connected for both full-game ATS and first-half ATS legs. The NFL totals portfolio also has true-close totals CLV connected for the tested totals windows. The season combined CLV is marked mixed because spread points and total points are different units and should not be blended casually.
Customer Framing
The results are shown the way a customer would naturally think about them. If a member risked $100 on every qualified NFL portfolio play, the page shows how many bets qualified, how many won, how many lost, how many pushed, how many units were earned, and how much money that equals at $100 per bet.
Important Note
The NFL backtest is evidence of how the model, Three Check process, and portfolio governor would have performed under the stated rules. It is not a guarantee of future profit. Sports markets change, injuries matter, pricing can move, and future seasons will not unfold exactly like the tested seasons. The value of the page is transparency: customers can see the method, the rules, the bet count, the wins and losses, and the historical financial result.