While most sports bettors are still focused on March Madness brackets and NBA playoff races, we’ve been quietly building something in the background that’s about to change the game.
The Lab is coming for baseball.
Opening Day is March 27. And when that first pitch crosses the plate, Donnie Dimes won’t be guessing which teams to bet on — the model will already know. Because we’ve been backtesting for months.
What We’ve Been Building
The MLB model isn’t a side project. It’s the same obsessive, data-driven approach that’s producing a 68% win rate across all sports right now — applied to the longest, most data-rich season in professional sports.
Here’s what’s under the hood:
- 4,795 games analyzed across the 2023 and 2024 seasons — every single regular season game, fully ingested and backtested
- 9,715 starting pitcher assignments mapped — because in baseball, the starter changes everything
- Park factors for all 30 MLB stadiums — Coors Field isn’t Petco Park, and the model knows exactly how each venue affects scoring
- Millions of parameter combinations tested — we’re not guessing at weights and thresholds, we’re sweeping through every possible configuration to find the optimal edge
The Architecture: How It Works
The MLB model uses a run-scoring projection system that’s fundamentally different from our basketball models — because baseball is a fundamentally different sport.
For every game, the model calculates:
- Team offensive output — runs per game, adjusted for the opponent’s pitching staff
- Starting pitcher adjustment — how much better or worse the actual starter is compared to the team’s average pitching. This is the single biggest variable in baseball betting.
- Park factor scaling — every stadium has a unique effect on scoring. The model applies park-specific multipliers to project accurate run totals.
- Home field advantage — smaller in baseball than basketball, but still measurable and still exploitable.
The result? A projected score for every game. When that projection disagrees with the sportsbook line, we have an edge.
The Backtest Numbers
We don’t launch models on vibes. We launch them on data. Here’s what the MLB backtest shows across nearly 5,000 games:
- 📊 Combined record: 1,638-1,511 (52%) — in a sport where 52% is profitable
- 💰 +1,588 units profit — massive volume, consistent edge
- 📈 +16.1% ROI — across all tiers of plays
- 🔥 ELITE tier: 53% win rate, +28.2% ROI — the high-confidence plays that drive the biggest profits
In baseball, you don’t need to win 60% of your bets. The moneyline market means you can be selective — taking plus-money underdogs when the model sees value, or laying reasonable juice on favorites the book is undervaluing. A 52-53% hit rate with smart line selection produces serious returns over a 162-game season.
Why MLB Is Perfect for AI Models
If you think basketball is good for quantitative models, baseball is even better. Here’s why:
The Season Is Massive
162 games per team. 2,430 total games per season. That’s not a sample size problem — that’s a goldmine. More games means more edges, more bets, and more compounding.
Starting Pitchers Create Exploitable Mismatches
Every day, sportsbooks have to price games based on who’s pitching. But the market isn’t always efficient at pricing the difference between a team’s ace and their fifth starter. That gap is where the model thrives.
The Market Is Softer Than You Think
Sharp money in baseball is concentrated on a few high-profile games each day. The rest of the slate? Underbaked lines that a good model can exploit consistently.
Totals Are a Second Edge
We’re not just building a moneyline/spread model. The totals model — projecting over/unders based on pitcher matchups, park factors, and tempo — is coming too. That’s two separate edges on every single game.
What This Means for You
When Opening Day arrives on March 27, Lab members will get:
- ⚾ Daily MLB picks with projected run lines, edges, and confidence tiers
- 📊 Starting pitcher analysis built into every prediction
- 🏟️ Park-adjusted projections that account for venue-specific scoring environments
- 💰 Calculated unit sizing — not just what to bet, but how much
- 📈 Full transparency — every pick tracked and graded on the results page
This is on top of the NCAAB, NBA, and NHL picks already running. The Lab is becoming a true year-round, multisport prediction engine.
The Bigger Picture
Most sports bettors treat each sport as a separate hobby. We treat it as a single problem: find mathematical edges the market hasn’t priced in, and exploit them systematically.
The same principles that produce our 68% all-sports win rate in basketball apply to baseball. Different inputs. Different variables. Same discipline. Same math. Same results.
March Madness is exciting. But while everyone else is filling out brackets with their hearts, we’re calibrating the model that’s going to run all summer long.
Ready for Opening Day?
The Lab is expanding. NCAAB. NBA. NHL. And now MLB. If you’ve been on the fence about joining, this is the time. Get in before the season starts and lock in access to every sport, every day, every edge.
162 games. 6 months. One model. Let’s get it.
JOIN THE LAB — MLB SEASON STARTS MARCH 27 →
Written by Donnie Dimes
AI-powered sports predictions. Every pick tracked. Every result graded. Learn more about Donnie →