Every pick on Bookie Bullies comes from a documented, reproducible AI model. No black box, no tout hype. Here's exactly what goes in, what comes out, and how picks are graded.
For every game on the slate, the model ingests sport-specific inputs and computes a closed-form win probability for each side. That probability is compared to the market moneyline. If the model's probability is higher than the market implies, the pick has positive expected value. Only picks that pass the strict gates are recommended. The card also shows reads for every game, including negative-EV ones. Recommended picks carry a Kelly-fraction stake; first-inning reads carry no stake.
Park, weather and umpire inputs feed the first-inning model, not the moneyline.
College football is no longer published: the model was retired on 2026-09-07 because it did not beat the market. While it ran we published leans, not picks, and the distinction was the whole point. A lean is what the model prefers. A pick is a lean that has proven it beats the closing line. Ours has not, so we label it honestly and grade every one in public at the college football board.
Five approaches have been tested against real closing lines. None beat the market:
Before it was retired, the leans were ranked in a deliberately counter-intuitive order: smaller disagreements outrank larger ones, and leans on favorites outrank leans on underdogs. Both rules come from results rather than taste. Games where we differ from the book by 14 or more points went 2-7 on the first graded slate, and the ridge shrinkage that produces our ratings structurally over-likes underdogs, so a favorite lean is one of the few reads not generated by that bias.
Sports betting models that apply the same equation to every sport leave money on the table because the underlying score distributions differ:
Raw model output is a margin or total estimate. That gets converted to a win probability with a correlated Skellam distribution. The model probability is then blended with the market using weights of 55% market, 30% model, 10% Statcast ensemble, and 5% latest posted line. When the model and market disagree by more than 5 points, the market weight rises. The blend shrinks overconfident predictions and avoids over-betting noisy inputs. The final blended probability is compared to the market moneyline. Expected value is calculated as (prob × decimal odds) minus 1. Only picks that pass the strict gates are recommended.
Confidence is independent of expected value. It comes from the calibrated win probability alone, not from edge size.
Every pick carries a unit-size suggestion based on a fractional-Kelly calculation (1/8-Kelly with a 0.25 standard deviation haircut, capped at 5 units) against the blended probability. The suggestion is informational. Users set their own bankroll rules. Bookie Bullies does not provide money management advice and is not a licensed financial adviser.
Every pick gets logged the moment it's published. Graded outcomes (W / L / P for push) appear at the first deploy after a game is final and graded, usually the 7 AM Pacific run the next morning. F5 picks are not graded. The running record is visible on the archive page. Nothing is hidden, edited, or retroactively removed.
Models are probabilistic. Even positive-EV picks lose, often in streaks. Variance is real and the sample size required to confirm edge is in the thousands of picks, not dozens. Models also miss real-world context: last-minute injury news, umpire substitutions, weather that shifts late. Treat the output as a starting point, not a command. Wager only what you can afford to lose. Gambling problem? Call 1-800-GAMBLER.