This paper defines thematic investing through transient correlations in residual returns. Using a novel bootstrap method, it finds that statistically significant themes exhibit trending, offering predictive insight into returns.
Interested in having your article published in the Financial Analysts Journal? Find out how.
Abstract
Thematic investing has grown in popularity even without a clear definition. We propose a risk-based definition of a theme and focus on themes that involve significant transient correlations of residual returns. We present a bootstrapping style approach to determine the statistical significance of the average pairwise correlation among stocks in a thematic basket. Analyzing thematic baskets provided by an investment bank, we find evidence of statistically significant correlations. The thematic baskets with statistically significant average pairwise correlation will have risk levels above predictions. Furthermore, they exhibit statistically significant trending. Baskets with insignificant average pairwise correlation do not trend on average.