Seasonal Autocorrelation and Hourly Clusters in Market Returns

Summary

The article examines whether apparent market randomness hides recurring hourly structure. It contrasts conventional autocorrelation of hourly price differences with seasonal autocorrelation, calculated after retaining observations from selected hours across days. It then uses hour-by-hour correlation heat maps to identify groups of hours whose returns move together, and scatter plots and related analysis to explore dependence between neighboring periods. The examples use EURUSD hourly data and report correlated hourly clusters, including groups around the early and middle parts of the day.

The author interprets the findings as evidence that market memory can reflect both recurring time-of-day effects and clustering among nearby increments. Later analysis connects these patterns to a trading model and optimization, but the article cautions that the EA is one possible interpretation and that optimization is only supporting evidence for patterns found statistically. The results are tied to the sampled instrument, period, lag choices, and analysis choices; the proposed relationship between neighboring increments remains a subject for further study.

Key ideas

• Conventional autocorrelation may miss relationships that emerge after grouping returns by time of day.
• Seasonal autocorrelation compares increments from recurring hourly slots across days.
• A correlation heat map can reveal clusters of hours with similar return behavior.
• The article attributes observed dependence to both seasonal recurrence and clustering of nearby increments.
• The findings and trading model are sample-specific, and the proposed explanation requires further validation.

Tags

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