Building a Rolling Sum from a Reusable Moving-Average Feature

Summary

This brief BigQuant forum post asks how to build one feature from another when the first feature is a moving-average comparison. The desired first feature assigns a positive value when the five-day average exceeds the ten-day average and a negative value otherwise. The second feature is intended to aggregate that signal over ten days to measure how often the shorter average has been above the longer one.

The author reports that the platform would not accept a rolling sum over the named feature and rejected an attempted integer conversion. The text ends with the question and does not include a reply or working solution. It therefore documents a feature-engineering problem rather than a resolved method. Readers can learn the intended signal construction and the specific obstacle encountered, but the post supplies no validation, code that works, or evidence about how the resulting indicator performs in a trading strategy.

Key ideas

• The question defines a signal from whether the five-day moving average exceeds the ten-day average.
• The intended second feature aggregates the first signal over a ten-day window.
• The author reports that summing the named feature and converting it to an integer both failed.
• The post contains no answer or working implementation, so the platform-specific issue remains unresolved.

Tags

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