Multivariate statistical analysis for portfolio selection of italian stock market

The use of bivariate cointegrated vector autoregressive models and Baba-Engle-Kraft-Kroner models ( Engle et al. 1995), is proposed for the selection of a stock portfolio (Markowitz type portfolio) based on estimates of average returns on shares and the volatility of share prices. The model put forward envisages the use of explicative variables. This article employs the intrinsic value of shares as a variable, which will make it possible to take the theory of value into account. The model put forward is applied to a series of data regarding the prices of 150 shares traded on the Italian stock market.