Lookahead Bias in Pre-Market Data Processing and Live Trading
Summary
This forum post raises a data-timing problem in a trading workflow. In the described backtest, a signal generated on one day is placed the next day, and pre-market processing can apparently retrieve that day’s opening and closing values through historical data. Such access may introduce lookahead bias if the strategy uses prices that would not yet have been available at the decision time. The author asks whether the same pre-market step runs in simulation and live trading, which date its final historical row represents, and how to use the opening price to decide whether to cancel an afternoon order.
The post offers no answers or implementation details; it documents an unresolved question rather than a validated procedure. Its useful lesson is to align each feature and order decision with information actually available at that timestamp, and to verify that backtest, simulation, and live data timing match. It does not establish how the platform behaves, so users would need platform-specific documentation or controlled checks before relying on the workflow.
Key ideas
- A backtest can be biased if pre-market logic reads prices from later in the same trading day.
- Data availability must be checked separately for backtests, simulations, and live execution.
- Historical data rows should be aligned to the timestamp when a signal or order decision is made.
- The post asks how to use the opening price for an afternoon order cancellation but does not provide a solution.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.