EARLY ACCESS

We're onboarding researchers gradually. Drop your email below and follow our progress on X.

Join Waitlist Follow on 𝕏

// HOW IT WORKS

01 / BACKTEST ANY IDEA
You type:
"Go long when funding rate is extremely negative and shorts are getting liquidated"
↓
10d runs the backtest:
Strategy: funding_mean_reversion
Signal: LONG on funding < -0.01% + short_liqs > $500k
Universe: BTC, ETH, SOL perpetuals
Period: 2023-01-01 to 2026-02-01
+243% Total PnL
77.6% Win Rate
3.46 Profit Factor
567 Trades
↓ same agent, opposite hypothesis
02 / KILL BAD IDEAS FAST
You ask:
"What about shorting when funding is extremely positive? Fade the crowded longs."
↓
10d runs the backtest:
Strategy: funding_mean_reversion_short
Signal: SHORT on funding > 0.03% + long_liqs > $500k
Universe: BTC, ETH, SOL perpetuals
Period: 2023-01-01 to 2026-02-01
-23% Total PnL
47.3% Win Rate
0.88 Profit Factor
312 Trades
Verdict: No edge. Do not trade.
Direction asymmetry detected: LONG on negative funding works.
SHORT on positive funding does not. Crowded longs don't capitulate reliably.

// RESEARCH

Every paper below was researched, written, and validated by an autonomous AI agent. It reads academic preprints. It suggests hypotheses. It runs the backtests. Humans review the output and learn alongside it.

Why Liquidation and OI Signals Don't Work: The Data Is Broken

Academic research shows exchanges systematically misreport open interest and delay liquidation data. We tested 40 hypotheses and found no edge.

#data-quality #liquidations #open-interest

Funding Rate Extremes: The Only Signal That Survived

After 40 hypotheses, one pattern held: LONG on extreme negative funding shows 77.6% win rate. Direction asymmetry is real.

#funding #mean-reversion #validated

// THE AGENT

It never stops working

While you sleep, it's scanning arXiv, SSRN, X, and Substack for new ideas. It's pulling live data from exchanges. It's testing another hypothesis. When something looks promising, it tells you. When something fails, it tells you why.

You stay in the loop

The agent suggests. You decide. Every hypothesis, every backtest, every conclusion surfaces to you for review. You learn what works, what doesn't, and why. The alpha is yours to act on.

Autonomous research

  • Scans arXiv, SSRN, X, Substack
  • Extracts tradeable hypotheses
  • Validates against historical data
  • Runs backtests unprompted
  • Reports findings. Waits for you.

Autonomous data

  • 19M+ candles and growing
  • Identifies missing datasets
  • Acquires what it needs
  • Integrates new sources on demand
  • Never blocked by data gaps

Get Early Access

Join the waitlist for private beta access.