Running LumiBot Backtests Across Historical Data Sources

Summary

This guide explains how to run a Python trading strategy backtest with LumiBot, choose a historical data provider, configure dates and sources, and review generated output. It describes ThetaData, Yahoo Finance, Polygon, custom Pandas data, and Polymarket CLOB history, noting differences in coverage and suitability. Historical data selection is separate from the broker used for paper or live trading, and an environment setting can override a data-source class selected in code.

The guide illustrates a simple daily stock strategy and explains that LumiBot produces a performance tear sheet, trade records, and indicator logs. These artifacts can help assess returns, drawdowns, benchmark comparisons, and trading behavior. The document is operational documentation rather than evidence about any strategy’s predictive value: it reports no comparative backtest results and gives no detailed treatment of data quality, transaction costs, or overfitting. Provider coverage and the example’s configuration should therefore be checked against a researcher’s asset, period, and testing needs.

Key ideas

• LumiBot can backtest strategies with vendor data, custom datasets, and Polymarket prediction-contract history.
• Historical data providers can be selected independently of the broker used for live or paper trading.
• Environment variables can supply backtest dates and override the data source configured in Python.
• Generated reports include performance charts and metrics, trade records, and indicator logs.
• The guide describes workflow and outputs but does not establish that a strategy will perform well live.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.