Backtrader

Cerebro coordinates a Backtrader simulation. Add a strategy, load a data feed with cerebro.adddata, run cerebro.run() and inspect the result with analyzers or cerebro.plot(). CSV files and pandas data provide alternatives to online feeds. The package is available through pip install backtrader; charting also needs Matplotlib.

The simulated broker is a useful place to examine assumptions. Its market orders normally execute at the next bar’s opening price. Limit and stop orders follow their own conditions, while setcommission and the slippage settings control trading-cost assumptions. Cash, current positions and portfolio value are tracked through the run.

By default, an order is filled completely in one step. A custom volume filler is needed to model partial execution. That matters for a study whose order size is large relative to available volume. Backtrader also supports resampling and replay, allowing one strategy to work with multiple timeframes.

The library itself is free to obtain and use under its GPL terms. Online data and live broker adapters have separate dependencies and access arrangements. The repository includes older connection examples, so a local file-based backtest is distinct from establishing that a particular live integration works with a current broker account.