QuantLab

Reproducible quantitative research and backtesting.

QuantLab turns a financial hypothesis into a reproducible, bias-aware experiment: download data, clean and validate it, build features and signals, run a vectorised backtest — with a delayed-execution barrier that prevents common look-ahead leakage — with configurable transaction-cost assumptions, measure performance and risk, validate out-of-sample, and generate an honest research report — all driven by one YAML config. Custom strategies remain responsible for their own causal feature and signal construction.

This project is for educational and research purposes only. It is not investment advice, and historical performance does not guarantee future results.

Where to start

• Architecture — how the modules fit together and why the pipeline is ordered the way it is.
• Data pipeline — sources, canonical schema, cleaning, validation, storage.
• Strategies — the strategy contract and how to add a new one.
• Strategy Explorer — the dashboard's interactive research/education mode for understanding each strategy.
• Backtesting — the accounting model and look-ahead-bias prevention.
• Validation — walk-forward, sensitivity, bootstrap, stress tests.
• Limitations — what this platform does not model, stated plainly.

Quick start

pip install -e ".[dev,dashboard,yahoo,extra]"
quantlab backtest --config configs/momentum_sp500.yaml
quantlab dashboard

Run quantlab --help for the complete command list and quantlab <command> --help for command-specific options.