Interpreting Cross-Asset Correlation Clusters
Summary
The document presents instrument groupings from a correlation-based clustering analysis of a broad futures universe. It compares solutions with different cluster counts, from two through ten, and interprets the resulting groups as risk-on, risk-off, volatility, and bond-related themes. As the number of clusters increases, broad groups split into narrower ones, such as separate equity, volatility, and bond clusters.
The examples show that clusters can mix asset classes: some groupings combine currencies, commodities, equities, and bonds, while others isolate regional volatility or particular bond markets. The author’s final hierarchy offers a qualitative way to think about these groupings, including core equities, Asian markets, commodities, and flight-to-safety assets. However, the document does not specify the correlation inputs, clustering algorithm, sample period, or validation method. The labels are interpretations of the displayed membership lists, not evidence that clusters are stable or predictive.
Key ideas
- Correlation clustering can group instruments across conventional asset-class boundaries.
- Increasing the requested cluster count splits broad groups into more specific regional and market themes.
- The displayed hierarchy interprets some clusters as risk-on, risk-off, or volatility-related.
- Cluster membership alone does not establish that the groupings are stable or useful for trading.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.