Black-litterman model with views prediction using elman recurrent neural network

Abstract

The Black-Litterman model is a portfolio model that considers investor views. The purpose of this study is to develop the Black-Litterman (BL) portfolio model with views prediction using Elman Recurrent Neural Network (ERNN) on LQ-45 stocks. The ERNN model is one of the neural network models that adjusts the input using the output feedback from the hidden layer. The BL portfolio is generated based on the capital assets pricing model (CAPM) excess return equilibrium. The data used in CAPM model must fulfill the normality assumption which is checked by using Jarque Bera test. The selected stocks for the portfolio are the member of LQ-45 stocks which meet the normality assumption and have the highest expected excess return CAPM value, those are AKRA, BBNI, INCO, and JSMR stocks. The ERNN model is employed to those stocks to obtain the views prediction. Then, the Black-Litterman portfolio is constructed by combining the ERNN views of the stock returns and the expected equilibrium return yielded by the capital assets pricing model. Three designs of relative views are considered, each design is distinguished from the percentage of each stock return prediction. The simulation shows that the best portfolio is constructed based on the design of the first views due to the most accurate views prediction.