NBA Predictions -- Moneyline, Spread & Totals Model Output

Model predictions for every game • 2025-26 Season

Error loading data

> Our NBA prediction model outputs moneyline win probability, spread cover lean, and over/under direction for every game on tonight's slate. Three model versions run in parallel -- V1 with the broadest selection, V2 with tighter thresholds, and V3 at the tightest confidence level. ML predictions have been above 60% in every season since 2020. Use these outputs as a research starting point alongside lineup news, rest data, and back-to-back schedules before placing a bet. Full season records are tracked in real time on the historical records page.

NFL 2025-26 Season Complete

47-38 | +100.5u

vs last season: +55.3u improvement ↑

Season begins October 2026

Model 1

Model 2

Model 3

Fave ATS

I'm a paragraph. Click here to add your own text and edit me. It's easy.

Fave ATS

I'm a paragraph. Click here to add your own text and edit me. It's easy.

Fave ATS

Today's Games • Apr 18

No games schedule today

MODEL 1

DET @ BOS

MODEL 2

DET @ BOS

MODEL 3

DET @ BOS

MODEL 1

DET @ BOS

MODEL 2

DET @ BOS

MODEL 1

DET @ BOS

MODEL 2

DET @ BOS

MODEL PERFORMANCE

MODEL 1

Sides

12-8

60%

MODEL 2

60%

MODEL 3

60%

MODEL 1

I'm a paragraph. Click here to add your own text and edit me. It's easy.

MODEL 2

60%

MODEL 1

60%

MODEL 1

I'm a paragraph. Click here to add your own text and edit me. It's easy.

MODEL 2

60%

MODEL 1

60%

Updated Apr 5, 7:12 am

How The Three Models Work Together

The core model calculates expected win probability by comparing each team's historical market line to tonight's posted number. When those diverge, that's the edge signal. V1 captures the wides set of those games each night -- the most selections, the most data. V2 narrows the threshold: fewer games, stronger signal. V3 is the tightest bucket, fewest plays, highest conviction. Looking at where all three models agree on the same side tells you more than any single model view alone.

What The Record Shows

V1 ML has ranged from 56% (2022) to 67.6% (2024) across six season. The spread market is harder to beat -- most seasons run in the 49-51% range, which is honest. The model's clearest signal is at the ML level, not the spread, which is useful before you use it, though 52% win rate on spread/totals is profitable. The full multi-year breakdown is on the historical records page.

Beta model expansion is planned for next season using the same architecture that will be trialed in WNBA in the summer.

NBA Predictions -- Moneyline, Spread & Totals Model Output

MODEL 1

MODEL 2

MODEL 3

CATEGORY

MODEL 1

MODEL 2

CATEGORY

MODEL 1

MODEL 2

MODEL PERFORMANCE

MODEL 1

MODEL 2

MODEL 3

MODEL 1

MODEL 2

MODEL 1

MODEL 1

MODEL 2

MODEL 1

How The Three Models Work Together

What The Record Shows