Asset Allocation and Intra-Sector Allocation Using Covariance and Precision Matrix Clustering

We investigate simply the usage of clustering method by inverse covariance estimation for asset allocation in finance. Allocation across various sectors of the market (i.e. sector ETF) is usually well understood through the usage of mean variance allocation. However, stocks inside a same sector (i.e. Biotechnology, IT,..) are little studied in the literature since high correlation between stocks leads to poor differentiation across stocks.Here, we apply covariance clustering method to study the behavior of one specific market sector: US listed biotechnology stocks. We show that this methodology allows differentiating highly correlated stocks and it provides more detailed information inside this sector. Additionally, the use of higher frequency data (intraday data) allows refining the clustering method and characterizing further each stock. This method can give further insight to select stocks inside a same sector