Educational/Not betting advice. For modeling and research purposes only.
GP
GametimePicks
v0.3
demo snapshot · sample slate

Transparent model leans on NBA player props.

GametimePicks compares model projections against sportsbook lines, surfaces edges with explanations, and tracks every result publicly. Educational analytics — not betting advice.

leans in snapshot
11
high-conf in snapshot
3
sample hit rate
54.8%
demo data
high-conf sample
63.3%
01

Compare projection to line

For each NBA player prop, the model produces a projected stat value and over/under probability. We pull the sportsbook line and convert the odds to an implied probability.

02

Quantify the edge

Edge = model probability minus implied probability. Positive edge means the model thinks the market is mispricing the prop. We surface only edges that clear a transparent threshold.

03

Track every result

Every lean is logged before tipoff and settled after the box score. Hit rate, calibration, and breakdown by market and confidence tier are all public.

demo data
This deployment is running on bundled demo data for offline development and previews. Configure ODDS_API_KEY in .env and run scripts/run_pipeline.sh to generate a live board.