Research On Asset Allocation Based On The Improved Black-Litterman Model

Securities investment funds have developed rapidly in years. However, in recent years, the stock market is much larger than before, because of the subprime crisis and the European debt crisis. In the case of high-risk, it’s an inevitable way to run a portfolio instead of one single stock. Related studies show that the efficient allocation of assets contributed93.6%investment performance. In recent years, the domestic securities investment funds begin to research quantification asset allocation models; most of them just copied those original models. However, China stock market is very different from foreign markets. Therefore, it is necessary to improve the original models, and make it be more adapting to China market.Black-litterman model is put forward by Goldman Sachs. It has been used in practical application from the beginning of its birth. After the development of these years, it has gained rapid approval. First of all, this paper makes a simple review on the classic mean-variance theory, and then the complex input parameters of Black-litterman model are discussed in detail. Using Bootstrap method and neural network model, original model has made the improvement. We use the Bootstrap method to do the Error correction; and neural network model can splendidly capture the complex variation in the stock market. Using the predictions of this model to replace views in original model may be able to enhance the performance.This paper selects41heavy weights from Shanghai50index excluded9stocks because of lack of data. With the error correction model, and using BP neural network prediction results as view returns vector, we get the expected returns and asset allocation results in next issue. In the first part of empirical study, the empirical results show that ANNs-BLR model is much more stable than mean-variance model, and the return and Sharpe ratio is both higher with or without short-selling constraints or weight limit. In the second part of empirical study, considering non-tradable shares existing generally in China stock market, this paper compared the performance of the original model with total market value weights, circulated market value and the improved model. The result shows that the circulated market value weights are more appropriate for the model, and can achieve better investment performance, and the adjusted model can also improve the performance.