Monday, February 22, 2010

Stock Market Sector Correlations as a Sentiment Measure

My recent post took a look at intermarket correlations and how those wax and wane with periods of relative fear and complacency in markets.

Above is a similar perspective, but now looking at intercorrelations among daily changes for four S&P 500 sectors: energy (XLE); financial (XLF); technology (XLK); and consumer discretionary (XLY). As before, I calculated the 10-day correlations among all of these; took the absolute values; and calculated an overall average correlation.

What we see is that the sectors tend to be highly correlated during periods of market decline and subsequent bounce. At relative market peaks, we tend to see divergences and patterns of differential sector allocation, which leads to lower correlations.

Once again, this is but a rough look, but a promising avenue for investigation. When traders and investors become fearful, there is a tendency to treat all stocks similarly, raising correlations. This is a useful sentiment gauge, as it tracks the actual portfolio-related behavior of money managers, not their stated beliefs about markets.