Try a popular strategy:
RSI + EMA Crossover
Bollinger Squeeze
MACD Divergence
Momentum Breakout
Triple Indicator
Copy-paste the strategy rules from the YouTube video description or transcript. The lab keeps the original idea intact, then tests nearby variants so Best PnL and Most Robust stay visibly separate.
Reading video description...
Claude Haiku extracting trading rules...
Generating 200 parameter variants...
Running on YOUR GPU (no servers)...
Computing robustness scores...
Ready
Stage 1
Original Guru Strategy
The video strategy is preserved as the baseline. The next steps evolve parameters around it instead of declaring the idea right or wrong.
Claude extraction
Stage 2
GPU Parameter Exploration
Strategy Arena generates nearby variants, then tests them locally in the browser. No trading rules leave the transparent result flow.
Starting GPU
No data sent to a black-box optimizer. Your browser runs the comparison and the methodology stays open.
Stage 3
Three lenses on the strategy
Best PnL is not automatically the winner. The robust variant is the one that survives drawdown, trade count and stability checks.
Grid pending
| Variant | Return | Sharpe | Max DD | Trades | Robustness |
|---|---|---|---|---|---|
| Original (Guru) | -- | -- | -- | -- | -- |
| Best PnL | -- | -- | -- | -- | -- |
| Most Robust | -- | -- | -- | -- | -- |
Original preserves the strategy as extracted from the video.
Best PnL can be fragile when a parameter set is too sharp.
Most Robust favors survivability over leaderboard drama.
Waiting for the evolution run...