QB1 Methodology & Model Limitations
QB1 Winning Edge is an independent NFL analytics platform. This page explains, at a useful level of detail, how QB1 produces a probability, how it compares that probability to market pricing, and where the model is weakest. Data flows in one direction only: NFL and market data feed the model, the model produces probability and Winning Edge, and the public pages present what the model already computed. Presentation never changes model output.
Elo Power Ratings
Every team carries a walk-forward Elo rating rebuilt from real final scores, where 25 Elo points equal roughly one point of spread. Ratings regress toward the mean between seasons so a strong prior year never overstates the current one. Ratings are computed in the order games were played, so a rating never sees a result from the future.
Game Simulation
Each matchup runs 10,000 simulated trials to produce win, spread, total and push probabilities rather than a single point estimate. That distribution is what makes uncertainty visible: a 58% favorite and a 78% favorite are treated very differently downstream.
Market-Implied Probability
Prediction-market contract prices and sportsbook moneylines are converted into probabilities, with two-sided prices de-vigged so the pair sums to 100%. Prices are also calibrated for market type and time remaining before expiry, because a price three weeks out behaves differently from a price at kickoff.
Market Blending & Fair Price
A weighted fair value blends the model probability, the calibrated market probability and sportsbook pricing. Blending is used to price the market, never to overwrite the model's own probability — the two numbers stay separately visible so users can see the disagreement rather than a smoothed average.
Data Quality & Confidence
Confidence reflects how much the model actually knows: games rated, freshness of the market snapshot, availability of injury and lineup information, and the width of the simulated distribution. Low data quality lowers the Winning Edge Score even when the raw probability gap looks large.
Edge Calculation
Edge is the difference between QB1 probability and market-implied probability, measured after transaction friction. Only differences that clear a minimum actionable threshold for that market type and time horizon are reported as opportunities; everything else is reported as PASS.
Market Movement & Timing
Stored price and line snapshots show how a market has moved since it opened, which reveals whether the modeled advantage is still available or has already been absorbed. Position-level entry zones and target exits are published to QB1 Pro members inside the app.
Model Limitations
QB1 does not know about unreported injuries, late lineup decisions, coaching intent, or in-game situational chaos. Ratings are less reliable early in a season and in the preseason, small samples are noisy, and a correct probability still loses frequently. Nothing on this site is financial advice, a guarantee, or a projection of profit.
Data Sources
Schedules, rosters and final scores come from established commercial and public NFL data providers; sportsbook pricing comes from an odds aggregation service; prediction-market pricing comes from public market data. QB1 does not connect to any brokerage or exchange account, hold trading credentials, or place orders.
Responsible Use
QB1 is analysis and tracking software for informational and educational purposes. Never risk money you need, never chase losses, and never treat a probability as a certainty. PASS is a valid and frequently correct output.
Audit the Model
Every pregame prediction is stored with a timestamp before kickoff and is never rewritten afterwards. See the graded track record for wins and losses side by side, or start with the current NFL prediction board.