vibe-trading

To backtest trading strategies from a CSV trade journal, you can use the Shadow Account loop to analyze your trades, distill actionable if-then rules with a minimum support of 3, and run multi-market backtests to measure performance. The system then renders an HTML or PDF report showing where your rules would have helped.

Factor analysis and Alpha Zoo benchmarking evaluate trading hypotheses against 452 pre-built alphas using qlib158, alpha101, gtja191, and academic models. This process converts investment ideas into measurable performance metrics through one-line CLI benchmarking and exportable manifests.

Yes, you can run multi-engine backtests across A-share, HK/US equities, futures, forex, crypto, and options. The system provides benchmark comparison panels to quickly assess strategy robustness across these different markets.

You need Python >=3.11 to run the MCP command entrypoint, while TUSHARE_TOKEN for A-shares and OPENAI_API_KEY for swarm teams are optional API keys. These optional keys enhance multi-agent research and A-share data access capabilities.

Yes, multi-agent research works with options pricing and behavioral diagnostics by applying swarm teams to strategy generation and robustness checks. This requires the optional OPENAI_API_KEY to fully enable the multi-agent research workflows.

Your shadow account backtest might fail to generate an HTML/PDF report if the extracted trading rules do not meet the minimum support of 3. Ensure your broker CSV trade journal contains enough data for the system to distill actionable if-then rules before rendering.