1 · The data layer
Contracts, picks, injuries, and transactions from public, timestamped sources. Nothing is presented as “real-time 100% accurate” — every cell shows when it was last validated and where it came from.
How it works
2 · The cap engine
A pure rule engine encoding the 2023 CBA (Cap Engine v2024-25.3). It returns a verdict and a reason code. It is versioned, and it must pass a suite of historical trades before any release ships.
3 · Trader AI
The AI reads need vectors computed from lineup data — not opinions — proposes packages, and writes the “why Team A says yes / why Team B says yes” analysis. Fit analysis is always labeled an AI-generated estimate.
4 · The value index
The Trader AI Value Index publishes its formula (production value − contract value + age + availability) and is labeled an index — never a market price.
What makes it different
- A second-apron validator with reason codes — the rule most trade content gets wrong.
- Proactive proposals — the site proposes trades; it doesn't only grade yours.
- Timestamped roster feed — “last checked + source + confidence” on everything.
- Show your work — formulas and rule citations in the open.
Data & licensing stance
The free site runs on public sources with attribution and rate limits. Advanced metrics (EPM, DARKO, LEBRON class) are licensed products and appear only under license. Commercial API access opens only on licensed data — scraped data is never sold.
Not affiliated with, endorsed by, or sponsored by the National Basketball Association. All team trademarks are property of their respective owners. Data for roster accuracy only, not betting advice. No insider information.