Diagnosing Empty Training Data in BigQuant Paper Trading

Summary

This support thread describes a BigQuant stock-ranking workflow that backtests successfully but fails when launched as simulated trading. The logs show that feature extraction runs, but later extraction steps return empty datasets; the ranker then receives an empty training set and stops with an error. The posted training-data configuration uses a historical date range, trading-day date binding, and a lookback period, while the live task is scheduled for a single bound date.

The evidence points to empty input data as the immediate cause, but the thread does not establish why extraction returned no rows or provide a confirmed fix. It therefore illustrates a useful debugging sequence: inspect extracted row counts and module inputs before treating the final model-training exception as the root cause. The issue may depend on the workflow’s data queries or live execution context; the excerpt is incomplete and should not be read as a general diagnosis for every backtest-to-live failure.

Key ideas

• The simulated-trading run fails after extraction steps produce empty datasets.
• The ranker’s immediate error is caused by an empty training input.
• Check intermediate extraction row counts to locate the failure upstream of model training.
• The thread does not confirm why the live task receives no rows or document a fix.

Tags

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