The aim of this paper is to test the
out-of-sample performance of the Black Litterman (BL) model for a German stock
portfolio compared to the traditional mean-variance optimized (MV) portfolio,
the German stock index DAX, a reference portfolio, and an equally weighted
portfolio. The BL model was developed as an alternative approach to portfolio
optimization many years ago and has gained attention in practical portfolio
management. However, in the literature, there are not many studies that analyze
the out-of-sample performance of the model in comparison to other asset
allocation strategies. The BL model combines implied returns and subjective
return forecasts. In this study, for each stock, sample means of historical
returns are employed as subjective return forecasts. The empirical analysis
shows that the BL portfolio performs significantly better than the DAX, the
reference portfolio and the equally weighted portfolio. However, overall, it is
slightly outperformed by the MV portfolio. Nevertheless, the BL portfolio may
be of greater interest to investors because -according to this study, where the
subjective return forecasts are based on historical returns of a rather long
past period of time-it could lead in most cases to lower absolute (normalized)
values for the stock weights and for all stocks to smaller fluctuations in the
(normalized) weights compared to the MV portfolio.
JEL classification numbers: C61, G11.
Keywords: Black-Litterman, Mean-variance, Portfolio optimization, Performance.