Method: these are model estimates, not guaranteed true probabilities. EST = built from the line, a positional average and game script only. PLAYER = also uses the player's recent average (last 5 games). DATA = also uses how the opposing defense has performed against that position. For NFL props both come automatically from free nflverse data, refreshed daily; college props stay EST unless you add columns 8-9. Each prop runs 100,000 simulated games. Hit % is the share of simulations on the pick's side of the line. Edge = hit % minus 52.4%, the break-even rate at -110. Every prop's baseline is adjusted for matchup: game script from the Sportsbook tab (the implied team score from the spread and total, which already reflects the opposing defense) plus any defense ratings you add. Props need matching team codes in the games list to get this. Pairs keep each player's own line: a QB's passing prop is matched with a teammate's receiving prop, and the two legs are simulated together (correlation 0.5). Both-hit % is the chance both legs land; pair edge = both-hit % minus 27.4%, the break-even for two -110 legs. Parlays tab: takes the strongest props on the board (by default only ones backed by a player's recent games, and never one the model rates above 72%, since that usually means an alt line or bad data) and tries every 2- and 3-leg combination that uses at least two different teams and one prop per player. All-legs-hit % comes from a joint simulation: legs in different games are independent, a QB and his own receiver are linked (0.5), and passing legs from opposite sides of one game are lightly linked (0.15). Break-even is 52.4% per leg at -110 (27.4% for two, 14.4% for three), or 33.3% / 20% on pick'em style payouts. Parlay edges compound any error in the single-leg percentages, so treat them as a shortlist to check, not a promise. Sportsbook tab: each game runs 100,000 simulations of the final margin and total. The centre is the market's own number (adjusted for lopsided juice) blended 85/15 with team ratings built from recent points scored and allowed (opponent-adjusted, from nflverse for the NFL and CollegeFootballData for college). Spread and total win % come straight from the simulation; moneyline win % starts from the no-vig moneyline. Edge = win % minus the break-even rate at that bet's actual odds. The game model is an assumption I have not validated yet: the Results tab tracks it separately under Game bets. Research only, not betting advice. 21+ where required; gamble responsibly.