⚠️ 100% SIMULATED PAPER TRADING — no real money is traded anywhere on this site. Educational & research project. Not financial advice.
Live experiment · simulated capital

Watch a machine
trade in the open.

An autonomous AI trading small-cap momentum on a simulated account — every position gone by the bell, every trade published.

See it trading How it works
EXPLORE

Live results — updated after each session · last build ET

VALIDATION WINDOW · trading days
VALIDATION WINDOW · completed trades

Equity curve (simulated $)

Daily P&L (simulated $)

Open positions

Today's candidate queue

Trade history — every closed trade, unedited

Incident log — we publish our failures too

All figures above are hypothetical, simulated results produced by a computer model with deliberately pessimistic fill assumptions and delayed market data. They are not real trades, involve no real money, and do not represent returns any person obtained or could have obtained. See the full Disclaimer.
The strategy · fully open

Five pillars. Every one required.

Small-cap "gap & go" momentum — the market's most dangerous regime, chosen because it punishes sloppy risk instantly. Published in full. Copy at your own risk.

01 / GAP
Up ≥10% today
Momentum already proven, not predicted. Preference ≥30%.
02 / VOLUME
≥5× average
Real crowds, not ghosts. Liquidity has to already be there.
03 / CATALYST
Real news
An identifiable catalyst — earnings, FDA, contracts — vetted by an LLM.
04 / PRICE
$2–$20
Room to move; high enough to clear the pennies.
05 / FLOAT
Under 10M shares
Scarcity is the fuel of outsized moves.
Risk rules · frozen in code

Limits the AI cannot touch.

RISK / TRADE1%
INITIAL STOP−5%
DAILY BREAKER−6%
FLAT BY15:55
SHOTS / TICKER1
FILLSHOSTILE
The scientific method

Pre-registered. Falsifiable.

The hypothesis and pass/fail gates were fixed before the first trade. 30 trading days. Three gates. Public failures.

GATE 01Mechanical

Zero critical failures — no naked stops, no duplicate orders.

GATE 02Risk fidelity

Not one trade breaches the 1% / stop / −6% rules. Verified by SQL, not trust.

GATE 03Economics

Expectancy ≥ +0.05R · profit factor ≥ 1.1 · drawdown ≤ 8%.

Anti-cheating commitments
  • Frozen rules: any change to sizing, stops, or the breaker resets the 30-day clock to zero.
  • Frozen learning: the self-tuning feedback loop is suspended for the window, so it cannot overfit mid-test.
  • Failures are data: software incidents are logged, classified, and published. A hidden failure is a failed experiment.
  • Falsifiable: if any gate fails, the result is NO-GO — as useful as a pass.
Honest limitations
  • Market data is ~15 minutes delayed; bid/ask spreads are synthesized. Real-time execution would differ.
  • Simulated orders can't move the market; real orders in micro-caps do.
  • One 30-day window is a small sample. A pass is evidence, not proof.