PreSnap Picks is powered by an ensemble prediction engine that combines multiple independent forecasting models. Each model analyzes a different dimension of NFL game data, including scoring trends, matchup dynamics, and statistical patterns. The final prediction is a calibrated blend of these perspectives.

Betting methodology: All P/L figures use a flat 1-unit bet per pick. No Kelly criterion, no odds-based sizing, no parlays. Flat betting isolates the model's raw predictive accuracy without any staking strategy inflating or deflating returns.

The Tier System

The model assigns moneyline and spread predictions a confidence tier based on how far its projected probability diverges from the market line. Over/Under picks use a binary value filter: picks either meet the edge threshold or they do not.

Best Bet: Highest conviction. The model identifies a significant edge versus the market line. (Moneyline & Spread only)

Bet: Solid edge detected. The model disagrees with the market enough to warrant a position. For O/U, this represents all picks that meet the edge threshold.

Low Value: Insufficient edge. The model does not see enough disagreement with the market to recommend a bet.

The model does not generate picks on every game. It targets specific matchups where it identifies a meaningful gap between its projection and the market price.

The following sections break down performance across three betting markets: Moneyline, Spread (ATS), and Over/Under.

Moneyline picks are straight-up winner selections. The model identifies games where the odds fall within a range that has historically produced mispriced lines, then assigns each pick a confidence tier based on how far the model's win probability diverges from the market's implied probability. When the edge is large enough, the pick is flagged as a Best Bet or Bet depending on the size of the disagreement.

Summary: 655 moneyline Best Bets & Bets over 5 seasons with a +11.14% ROI. Best Bets hit at 60.6% with a +17.08% ROI.

The model generates its own predicted spread for each game and compares it against the closing market line. When the disagreement is large enough, the pick is flagged as a Best Bet or Bet depending on the size of the edge.

Key takeaway: The break-even rate for -110 spread bets is 52.4%. The model's value spread picks hit at 57.4%, well above break-even, generating a +9.48% ROI. Best Bets are the standout at 58.9% with +12.59% ROI.

The model predicts a total score for each game and compares it against the market line. When its prediction diverges meaningfully from the posted total, the game is flagged as a value O/U pick. Picks are classified as Bet when the edge falls within a productive range, or Low Value when the edge is too narrow or falls outside the model's reliable range.

Over vs. Under: Overs hit at 56.1% (+6.87% ROI) across 496 picks. Unders hit at 59.0% (+12.62% ROI) across 486 picks. Combined O/U Best Bets & Bets have produced +95.45u in profit.

This section combines every Best Bet and Bet from Moneyline, Spread ATS, and Over/Under into a single portfolio view. It answers the question: what is the total return if a bettor followed all of the model's picks?

Portfolio summary: Across 2430 total Best Bets & Bets over 5 seasons, the combined portfolio has produced +243.57u in profit at a +10.02% ROI.

5 of 5 seasons were profitable across the combined portfolio. The table below shows how Best Bets & Bets performed in each individual season.

We’re excited to be rolling out the new models for each three markets for the NFL Season this year, our value picks also update as the odds change for the games, so no matter when you check the picks you’ll always have a real-time edge with NFL Games.