Diagnosing Empty Prediction Data in Simulated Trading
Summary
A BigQuant user reports that a simulated run fails at a missing-data cleanup step, even though the backtest sample contains two rows. The discussion distinguishes the historical backtest data from the prediction data used during simulation, which may be empty or contain rows that are all removed as missing. A successful resolution traced the problem to prediction data containing Beijing Stock Exchange securities that the cleanup step discarded.
Suggested diagnostics are to align the backtest start and end dates with the current day to check whether data exists for the prediction period, and to print or inspect the prediction dataset before cleanup. The account is a troubleshooting anecdote rather than a general guide to platform behavior; it offers no systematic explanation of the filtering rules or broader validation procedure.
Key ideas
- Simulation can fail at missing-data cleanup even when the historical backtest has rows.
- Inspect the prediction dataset separately from the backtest sample.
- Check whether data exists for the current prediction date by aligning the backtest dates with that day.
- In the reported case, prediction rows for Beijing Stock Exchange securities were removed, leaving no data.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.