Auto-Quant

What this workspace does

Spawns a workspace as a local Auto-Quant clone on its own autoresearch/<tag> branch. The agent runs the existing prepare / backtest / optimize pipeline, edits strategies under user_data/strategies/, logs results to results.tsv, and commits as it goes — so every iteration leaves an auditable git trail.

When to spawn this

• ›You want to test a new strategy idea (mean-reversion, momentum, factor combinations) and let the agent iterate on parameters.
• ›You're exploring how to combine multiple indicators against Auto-Quant's existing data layout.
• ›You want a backtest loop that runs alongside other workspaces without sharing data directories.

Isolation

Each workspace gets its own user_data/data/. First run, the agent runs uv run prepare.py to fetch OHLCV from Binance. The isolation is deliberate: different runs may want different timeframes or symbols, and a shared cache would silently mix incompatible files across workspace generations.

What you'll see in Inbox

• ›Backtest result summaries (Sharpe, drawdown, key periods) as the agent finishes runs.
• ›Strategy diff notes when the agent commits a meaningful iteration.