How To Use
What this app does
JonRon Model predicts the probability that an MLB batter will hit a home run in today's game using a v4 stacked ensemble of 4 machine learning algorithms trained on 301,440 plate appearances. It compares those probabilities against sportsbook odds to identify bets with positive expected value (+EV), and provides pitch-level matchup analysis so you can verify the edge before betting.
Home Run Probabilities
The main page shows every batter-pitcher matchup for today's slate, ranked by home run probability. Here's what each column means:
Tip: Use "Show +EV only" to filter to just the bets with an edge. Sort by +EV to find the strongest edges. Filter by team logo to focus on a specific game.
Barrel Edge
This tab ranks batters by their barrel rate against the specific pitches they'll face today. Unlike the aggregate HR probability, Barrel Edge shows you the raw contact quality data behind each matchup.
How to use it: A batter ranked #15 in HR probability but #1 in Barrel Edge is someone the aggregate model may undervalue. If they barrel 30%+ of the pitcher's primary fastball thrown 50% of the time, that's a strong signal. Click any row to expand per-pitch detail with barrel%, FB%, exit velo, hard hit%, pitcher barrel allowed, sample size, and date range.
Placing bets
Click the "Bet" button on any prediction to open the bet modal. As you type in your odds, the modal calculates your expected value in real time, so you can instantly see if a line is still +EV at your book's price.
Matchup detail page
Click any batter on the daily board to open the full matchup breakdown. This is where you do due diligence before placing a bet.
What makes a strong HR bet
Red flags
v4 Model (2026 Season)
The current model is a stacked ensemble of XGBoost (Tweedie loss), LightGBM, CatBoost, and EBM (Explainable Boosting Machine) trained on 301,440 plate appearances from 2024-2025 with 68 features.
New in v4: Pitch-mix weighted features account for how a batter performs against the specific pitches a pitcher actually throws, not just aggregate stats. This catches matchups where a batter's overall numbers look good but they struggle against the pitcher's primary arsenal.
Calibration: Probabilities are calibrated so that when the model says 5%, roughly 5% of those predictions result in a HR. A 10% per-plate-appearance cap prevents physically impossible game-level probabilities. The highest single-game probability is capped around 40%, which represents an elite hitter in an ideal matchup with 4-5 plate appearances.
Weather: Every prediction includes game-time temperature, wind direction, and air density. Coverage is near 100% for all MLB parks.
Top features: Platoon advantage, batter total PA, hard hit rate, pitcher velocity, launch angle, park HR factor, pull rate, and pitch-mix weighted ISO are the most predictive features in the model.
Performance tracker
Tracks your betting results and the model's overall prediction accuracy.
Model performance shows hit rates by probability tier and +EV range across ALL predictions, not just your bets. This validates the model is calibrated.
Your bet performance tracks W-L, net units, and ROI on bets you've actually placed. Each bet shows the EV at the time you placed it.
Expect variance. You need 200+ bets before results are statistically distinguishable from luck. Focus on process (betting +EV spots) not outcomes.
Same-Game Parlays
Build custom HR parlays by searching players and adding them to your slip. The builder shows combined probability, fair odds, and whether the book's parlay price offers +EV. Model-generated parlays surface the highest-probability and highest-EV combinations automatically. Note that books apply correlation adjustments to SGP pricing, so model fair odds assume independent events.
Bullpen Rankings
All 30 teams ranked by bullpen HR vulnerability using a composite score of barrel rate, fly ball%, hard hit%, HR/9, and inverse whiff%. High-vulnerability bullpens are targets for late-game HR bets and same-game parlays. Sortable by any stat column. Hover any header for the stat definition.
Bankroll management
1 unit = 1% of your total bankroll. If you have $1,000, 1 unit is $10.
Most bets will be 0.5-1.5 units. The model uses quarter-Kelly criterion for conservative sizing.
Never chase losses. Home runs are rare events (3-4% of plate appearances). Even elite hitters go multiple games without one. This app gives you a mathematical edge, not certainty.
Important: Bet responsibly and within your means. A minimum bankroll of 100-200 units is recommended. Expect 500+ bets to confidently distinguish skill from variance.