Run a strategy¶
For a Python script or notebook, construct Backtest(data, MyStrategy, ...) and call run(...). For a reusable strategy file carrying default settings, declare @backtest(...) and use runtime.run(namespace, ...).
Choose an entry point¶
| Goal | Entry point |
|---|---|
| One run with explicit settings | Backtest.run |
| Apply defaults from a decorated strategy | runtime.run |
| Find a best parameter combination | Backtest.optimize |
| Save and query every candidate | runtime.optimize |
Both run paths return a pandas Series with statistics and execution tables. Every run gets fresh strategy/account state. runtime requires a loaded module namespace, aware market-data index, DataContext, and InstrumentSpec; the direct Backtest path can omit context.
Keep units explicit¶
Backtest(commission=0.002) and @backtest(commission_value=0.2) both mean 0.2% fees, but use different argument units. buy(percent=20) means an incremental 20% budget; order_target_percent(target=0.2) means a 20% target position. Read the relevant parameter table before transferring values between APIs.
Full Backtest parameters · Decorator and runtime example · Results