Portfolio construction and empirical testing of Black Litterman model

Portfolio construction and empirical testing of Black
Litterman model
Gianpaolo D’Orazio
Abstract
The main subject of this thesis is portfolio construction and Black-Litterman estimation model. In the
first chapter, I will analyse gradually all the techniques which are necessary to build a reasonable
portfolio. The starting point is asset allocation; it follows risk and correlation matrix, then
Markowitz’s mean-variance optimization with its weaknesses and strengths.
The unreason ability of simple Markowitz’s optimization will be shown and the model will be first
improved by intragroup constraints, then by a resampling technique, then by a Bayesian estimation
method, the Black-Litterman model.
The second chapter deals with calibrating the Black-Litterman model I order to overcome the
problems arisen by the difficulty in its parameters’ estimation.
Finally, Black-Litterman estimation model will be compared with resampling and afterwards with
infragroup constraint technique in order to elect the best model available.
Key words: Strategic asset allocation, risk and correlation measures, portfolio optimization,
Black-Litterman model, resampling technique, infragroup constraints