Live Paper-Trading Proof
Simulated executions on live Binance prices with real fees. No real capital currently at risk (real-capital lessons below). We run the worst strategies in paper to prove we don't hide losses. Updated daily. Nothing is curated.
123960 trades • 117 days of data • 48coins • Last trade: —
Trust the pipeline? Unlock the survivor library — the strategies that cleared every gate above.
$10 lifetime →Real capital lessons
The paper story is not the whole story. In May we deployed real Kraken capital. Here is what happened.
Why it happened. The live-trading engine wrote its own peak and drawdown counters, and both were broken from day one — peak stayed at $0 while the account bled. The portfolio watchdog was pointed at paper components only; the real Kraken state file was never wired into the alerting layer. Six weeks of silent bleeding invisible to every automated check.
What changed. Live trading was paused via kill-switch on 2026-06-16. The portfolio aggregator was rewritten to (a) surface live-trading-state.json alongside paper components, (b) recompute peak and drawdown from closed-trade cumulative P&L instead of trusting the broken counter, (c) fire a critical alert whenever peak=0 with a moved balance. A uniform last-activity watchdog was added so any silent-bleed pattern surfaces within days rather than weeks.
When it resumes. No real capital re-deploys until: (1) the watchdog has been verified working end-to-end via a test alert, (2) 90+ days of clean paper forward-test data on the strategy that would trade, (3) the peak math verified against a synthetic loss scenario. The -26.85% is the whole data point — no more capital is added to test whether it was luck.
This section exists because the honesty framing everywhere else on this page has to be earned. Anyone can publish paper results; publishing a real loss with the exact watchdog failure that missed it is what "we publish our losses" is supposed to mean.
Pipeline: Research → Live Capital
Every step public. Every number real. Every commitment written in advance.
Our commitment (written in advance)
- A gene must pass the research-lab approval gate before forward testing. The gate has three lanes: standard (positive walk-forward P&L, 15+ trades, 30%+ WR, Deflated Sharpe fluke probability < 50%), near-breakeven with strong fundamentals (100+ trades, 60%+ WR, fluke < 50%), or high-conviction selective (8+ trades, 75%+ WR, clear positive P&L). Tightened 2026-04-30 after a gate audit retired 549 prior approvals as noise.
- Forward paper trades must match backtest predictions within 15% over at least 30 live trades before any real capital is considered.
- When a gene is cleared for live capital: initial allocation $500-$1,000 per strategy, max 5 strategies simultaneously = $5,000 total real capital exposure for the first 60 days.
- Every real-capital trade will appear on this page with entry price, exit price, fees paid, slippage measured. Wins AND losses.
- If real-capital P&L diverges more than 20% from ongoing paper prediction, the strategy is auto-retired and the divergence analyzed publicly.
Nobody else publishes the criteria for deploying real capital BEFORE they deploy it. That's the point. When we do put real money in, you'll see the decision and its outcome in the same place.
Baseline vs. Research filter
Live paper trading, same data, same fees, same period. Research filter has much less sample. Numbers firm up around 100 trades.
Hypothetical · forward test · 2%-risk simulation on $1,000
Same Binance data, same real fees, same paper-trading window. The research-filter portfolio is still a small sample but points in a different direction than the baseline. Both run live. Both update daily. If research goes negative tomorrow, you'll see it here.
The Score (live paper trading)
Equity Curve (2% risk per trade)
Hypothetical · forward test · $0 capitalThe Why Behind Each Trade
v2 adaptive engineEvery trade the research engine takes comes from a gene: a specific signal, regime gate, and fee-baked exit. Here's the last 15, in plain English.
Daily trades, strategy mix, per-timeframe performance, methodology.
This is real data from live Binance trades. Not cherry-picked. Not backtested. We show the wins and the losses. The system auto-updates daily.
If our strategies lose money, you'll see it here first. That's the point.
Test Your Own StrategyNot financial advice. Simulated paper trades on Binance data. Past performance does not guarantee future results. 2% risk per trade simulation.