The Lab doesn’t sleep. While most tout services are recycling the same tired narratives, we’ve been deep in the data — rebuilding, backtesting, and stress-testing until the numbers tell us we’re ready.
Today I’m rolling out two major updates. Let’s get into it.
NCAAB Model: V8.1 → V9.2
The college basketball model just got a serious overhaul. We didn’t slap a fresh coat of paint on V8.1 and call it a day — V9.2 is a structural rebuild with three key upgrades:
Enhanced Quality Filters — The model now applies tighter screens on which games even qualify for analysis. Not every matchup deserves your money. V9.2 knows the difference.
Combined Structural + Projection Gate — Instead of running structural analysis and projection-based filtering as separate steps, V9.2 fuses them into a single gate. A pick has to pass both checkpoints simultaneously. (New to spread betting? Check out our complete guide to ATS betting.) This eliminates the “looks good on paper but falls apart in practice” problem.
Improved Selectivity — This is the big one. The model is being extremely picky. Fewer picks, higher conviction. We’re not spraying 8-10 plays a day and hoping for volume — that’s how 65% of bettors lose. We’re waiting for real edges and striking when the data says so.
We reset the season record on February 19 — Day 1 for V9.2. Since then: 5-2, +9.60 units. Small sample, but the early signal is strong. The model is finding value exactly where it should be.
NBA V2: The Launch
This one’s been cooking for a while.
After extensive backtesting across 513+ games — effectively the full 2025-26 NBA season to date — the NBA V2 model is officially live for both ATS and Totals.
Here’s what the backtest showed:
ATS Performance
- 61.3% win rate on selective picks
- +22.6% ROI
- The model identifies mispriced spreads using adjusted offensive and defensive ratings, pace projections, home court advantage modeling, and rest day analysis.
Totals Performance
- Up to 85% win rate on high-edge plays
- The totals model keys on pace differentials, defensive efficiency matchups, and situational factors that the market consistently underweights.
Before going fully live, we ran the model in ghost mode — tracking picks without publishing them. It went 2-0 in ghost testing, confirming the backtest wasn’t just noise.
What Powers It
The NBA V2 engine runs on:
- Adjusted offensive/defensive ratings — not raw stats, but efficiency numbers adjusted for opponent strength and pace
- Pace projections — how fast or slow a game will actually play, not just team averages
- Home court advantage modeling — because not all home courts are created equal
- Rest day analysis — back-to-backs, travel, and schedule density matter more than most bettors think
This isn’t a “feel” model. Every pick has a quantified edge before it goes out. That’s why AI is the only real edge left in modern sports betting.
What This Means for You
The Lab is now running two upgraded engines — NCAAB V9.2 and NBA V2 — both designed around the same philosophy: be selective, find real edges, let the math do the work.
No fluff. No filler picks to look busy. Just data-driven plays with transparent tracking — and we give it away free.
The numbers are the numbers. And right now, the numbers are speaking. 🎯
🔬 Join The Lab — Get the Picks
Written by Donnie Dimes
AI-powered sports predictions. Every pick tracked. Every result graded. Learn more about Donnie →