Unpriced Lab · AI trading research

A small trading lab built on frontier AI

We use frontier models to build simulated trading agents and manage paper portfolios.

Small positions. Public results.

Loading…

What we do

We turn public trading ideas into code we can test

01 / DISTILL

Read what traders publish

Letters, livestreams, books and regulatory filings. Frontier models help us turn each stated method into explicit entry, exit and sizing rules.

02 / SIMULATE

Run every strategy the same way

Each one gets a $1M paper portfolio. Decisions use the close, fills happen at the next open, and every trade pays costs.

03 / PUBLISH

Show the work, including losses

Rules, holdings, filings and returns are public. Strategies that lose money stay on the board.

Trader Lab

Seventy portfolios in three groups. One million paper dollars each.

Jev scored 156 funds, independent investors and members of Congress. We kept the 70 we could simulate honestly and mirror their filings, run their published rules or replay their disclosed trades.

$70M Paper capital under simulation
# Trader Method Return Max DD
Loading…

S&P 500 heatmap of the latest archived session

S&P 500 heatmap

Market data

A daily map of the U.S. market

After every completed session we rebuild the S&P 500 heatmap, write a short AI-assisted summary and archive it by date.

— Sessions in the Daily Archive

We are not a fund

Unpriced doesn't take outside money, manage accounts or give investment advice. We publish what we test so anyone can check it.