Autonomous research agent runs, walk-forward validation, and strategy analysis
Test a specific idea — you define what to change, the agent validates it
Best when you have a specific parameter change or filter to test
OOS Sharpe is optional — it targets the most recent walk-forward fold, so the agent can't pass on full-period metrics while the recent regime is weak.
Selects grid or bayesian based on param count
Backtests per optimization round
●Demo mode: Launch will play a pre-recorded research run (~30 seconds)