Opening Range Breakout with AI

Code Lumibot

Summary

This strategy scans a fixed universe of large, liquid US-listed stocks for breaks above the high established during the first 15 minutes of the regular session. A research agent identifies and ranks stocks that have since closed above that level; a separate trading agent can act on the research, with checks run hourly after the opening range is complete.

The trading rules allow at most one position. The opening-range low defines the stop, position size is set so that a stop-out risks no more than 1% of account value, and the profit target is 1.5 times that risk. The position is also closed if price falls back into the range and must be closed before the session ends. The code supports backtesting and live operation, but gives no performance results, execution details, or evidence that the agent ranking improves selection. The fixed universe and hourly review schedule also constrain how broadly or quickly the method can respond.

Key ideas

The strategy defines its opening range using the first 15 minutes of the session.
A research agent ranks stocks that close above their opening-range high.
A trading agent takes at most one breakout position and uses the range low as its stop.
Position size limits the planned stop risk to 1% of the account, with a target at 1.5 times that risk.
The strategy exits on a return into the range or before the close, but reports no performance evidence.

Tags

Equities Breakout Position sizing Risk management Execution US markets

Full text

# ai_opening_range_breakout.py


```py
"""Opening Range Breakout AI Trading Bot.

Marks each stock's high and low from the first 15 minutes of the day, then buys
the stock that breaks out above that high. A research agent scans a list of big
stocks every hour. A trading agent buys the best breakout with a profit target
and a stop, and is out of the market by the close.
"""

from lumibot.strategies import Strategy


class AIOpeningRangeBreakoutStrategy(Strategy):
    parameters = {"universe": ["SPY", "QQQ", "AAPL", "MSFT", "NVDA", "AMZN", "META", "GOOGL", "TSLA", "AMD"]}

    def initialize(self):
        self.sleeptime = "1H"
        self.agents.create(
            name="researcher",
            allow_trading=False,
            system_prompt=(
                "For each stock in the universe, find today's opening range: the high and low from 9:30 to "
                "9:45 ET. List the stocks that have since closed above their opening range high, best first. "
                "Do not trade."
            ),
        )
        self.agents.create(
            name="trader",
            allow_trading=True,
            system_prompt=(
                "Hold at most one stock. If we hold none, buy the best breakout. The stop is the opening "
                "range low: size the trade so the stop loses at most 1% of the account, and place a profit "
                "target at 1.5 times that risk. Sell if the price falls back into the range, and always "
                "before the close."
            ),
        )

    def on_trading_iteration(self):
        if self.get_datetime().hour < 10:  # the opening range is not finished yet
            return
        facts = {"universe": self.parameters["universe"]}
        research = self.agents["researcher"].run(task_prompt="Find the best breakout.", context=facts)
        self.agents["trader"].run(task_prompt="Trade the breakout.", context={**facts, "research": research.summary})


if __name__ == "__main__":
    from lumibot.credentials import IS_BACKTESTING

    if IS_BACKTESTING:
        from lumibot.backtesting import AlpacaBacktesting

        AIOpeningRangeBreakoutStrategy.backtest(AlpacaBacktesting)
    else:
        AIOpeningRangeBreakoutStrategy().run_live()

```

Shown in full with attribution under the source's licence. Licence: GPL-3.0

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

AI-Assisted Opening Range Breakout with Risk-Based Position Sizing

Summary

Key ideas

Tags