Strategy API Overview¶

Most LumiBot strategies need one lifecycle method and a small group of data, account, and order methods. Start with this complete daily-stock example:

from datetime import datetime

from lumibot.backtesting import YahooDataBacktesting
from lumibot.strategies import Strategy


class BuyAndHold(Strategy):
    def initialize(self):
        self.sleeptime = "1D"

    def on_trading_iteration(self):
        if self.first_iteration:
            price = self.get_last_price("SPY")
            quantity = int(self.get_cash() // price)
            order = self.create_order("SPY", quantity, "buy")
            self.submit_order(order)


if __name__ == "__main__":
    BuyAndHold.run_backtest(
        YahooDataBacktesting,
        datetime(2025, 1, 1),
        datetime(2025, 2, 1),
    )

Core methods¶

Common Strategy methods¶
Task	Method	Detailed reference
Read the latest price	`self.get_last_price(asset)`	Data
Read historical bars	`self.get_historical_prices(asset, length, timestep)`	Data
Read cash and positions	`self.get_cash()` and `self.get_positions()`	Account Management
Create and submit an order	`self.create_order(...)` and `self.submit_order(order)`	Order Management
Run logic on a schedule	`initialize()` and `on_trading_iteration()`	Lifecycle Methods
Backtest the strategy	`Strategy.run_backtest(...)`	Backtesting Function

Use Strategy Properties for fields such as first_iteration and portfolio_value. Continue to Strategy Methods for the complete categorized method reference or Run your first AI backtest to add an AI agent to the same lifecycle.

Choose how to use LumiBot¶

Use a Strategy subclass for trading and backtesting, including AI teams. Use Use LumiBot in another Python project for selected data/research helpers in another program. A configured strategy can run continuously with run_live() or execute one scheduled lifecycle with run_live(run_once=True). The latter is not an arbitrary step-through-backtest API.

There is no requirement for a function named main. An if __name__ == "__main__": guard prevents your runner executing on import. See LumiBot for coding agents for exact entry points for coding agents.