What the models do, what they don't, and how picks are chosen.
Every simulation starts from the market. The posted spread sets the median margin, a small weight goes to our ratings (fit to real results), and the shape of the outcome distribution comes from decades of real games. We then simulate 10,000 games and report win probability, fair moneyline, cover/push/over/under probabilities and the estimated value at the book's price.
We show real opening versus current lines, price (juice) moves, moneyline moves and key-number crossings from DraftKings, plus injuries, kickoff weather and, when a splits feed is connected, ticket vs money splits and reverse line movement. Reported facts and inferred signals are labeled separately.
A pick must clear thresholds for validated edge, data quality and uncertainty. Each lists book, line, odds, publication time, reasoning and the worst price still recommended. Many days have none.