An AI model that finds where the MLB market misprices probability — validated on 9,400+ backtested underdog bets (2018–2025), stress-tested through 120 scenario simulations, and posting every pick publicly since April 2026. It tells you what to bet, how much to bet, and exactly why.
Free 2-minute assessment · No credit card
Backtest, 2018–2025 seasons · live 2026 record further down this page
You're betting into a market that already priced in everything you know. Intuition doesn't beat algorithms.
$200 on a 2% edge, $50 on a 7% edge. Most bettors size on conviction, not math. The Kelly criterion exists for a reason.
Without a system, a bad week becomes a bad month. You chase, you oversize, you abandon discipline. Math compounds against you.
Custom rating system updated after every game. Home advantage, margin of victory, between-season regression.
Base ProbabilityTeam batting differential, regressed toward league average and normalized to z-score. The heaviest-weighted factor in the live model.
25% WeightFIP-based z-scores for each starter, regressed by innings pitched. Where the market makes its most predictable mistakes.
20% WeightRecent reliever workload and rest differential. Fresh bullpen = real, exploitable edge the market underweights.
6% WeightCoors Field (114) vs. Dodger Stadium (96) is an 18-point swing. The model adjusts for where the game is played.
6% WeightCross-country red-eye into a day game? Quantifiable disadvantage. The model sees what your eyes miss.
4% WeightThis is a real pick from our live record — not a vague "lean" or a lock emoji on Twitter. You see the exact model probability, the market's implied probability, the calculated edge, and the mathematically optimal bet size for your bankroll.
The full factor breakdown shows you why the model likes the pick. Is it a pitching mismatch? A tired bullpen? An Elo gap the books haven't priced? You decide whether to follow.
Anyone can show a pretty backtest. We went further — running every possible ordering of five MLB seasons (all 120 permutations) to answer the question most systems avoid:
Every single permutation ends profitable. Even the worst-case scenario — where the toughest seasons hit back-to-back — turns $10K into $57K.
| Typical Tout | PICKSJUICE | |
|---|---|---|
| Methodology | "Trust me bro" | 6-factor logistic regression |
| Edge quantified? | ✕ Star ratings | ✓ Exact % on every pick |
| Bet sizing | ✕ "3-unit play" | ✓ Kelly criterion |
| Transparency | ✕ Black box | ✓ Every factor visible |
| Track record | ✕ Self-reported, cherry-picked | ✓ Every pick public, updated daily |
| Risk management | ✕ None | ✓ Floor protection, dynamic caps |
| Losing picks | ✕ Deleted or blamed on luck | ✓ On the record, forever |
Current model, live since June 13, 2026. These numbers load straight from the same database that logs every pick — they update themselves, win or lose.
2-minute strategy assessment. We match you with the right sizing approach and risk profile — then show you exactly what the system has done.
START THE ASSESSMENT →