Tactical asset allocation: an artificial neural network based model

An artificial neural network was trained to support a tactical asset allocation investment strategy. The allocation strategy considers three asset classes: US stocks, bonds and money market. The neural network was trained to forecast the probability that each asset class would outperform the other two by the end of a one-month period. The neural network was trained with the backpropagation algorithm. A tactical asset allocation portfolio was invested in the asset class expected to have the best performance according to the neural network prediction. The strategy was simulated during a one-year period. During the simulation period the strategy outperformed the S&P500 Index by 1,792 basis points. The artificial neural network prediction was accurate 92% of the time.

[1]  Thomas G. Dietterich.  Overfitting and undercomputing in machine learning , 1995, CSUR.

[2]  W. Sharpe.  Asset allocation , 1992 .

[3]  Donald C. Wunsch,et al.  Advanced neural network training methods for low false alarm stock trend prediction , 1996, Proceedings of International Conference on Neural Networks (ICNN'96).

[4]  Sid Browne,et al.  The return on investment from proportional portfolio strategies , 1998, Advances in Applied Probability.

[5]  Jung-Hua Wang,et al.  Stock market trend prediction using ARIMA-based neural networks , 1996, Proceedings of International Conference on Neural Networks (ICNN'96).

[6]  Stephan Rudolph,et al.  On topology, size and generalization of non-linear feed-forward neural networks , 1997, Neurocomputing.

[7]  Vijay S. Desai,et al.  The Efficacy of Neural Networks in Predicting Returns on Stock and Bond Indices , 1998 .

[8]  Yoshua Bengio,et al.  Experiments on the Application of IOHMMs to Model Financial Returns Series * , 2002 .

[9]  W. N. Street,et al.  Financial Asset-Pricing Theory and Stochastic Programming Models for Asset/ Liability Management: a Synthesis , 1996 .

[10]  Xu Wenhua,et al.  Training neural network with genetic algorithms for forecasting the stock price index , 1997, 1997 IEEE International Conference on Intelligent Processing Systems (Cat. No.97TH8335).

[11]  Hailiang Yang,et al.  Asset allocation with time variation in expected returns , 1997 .

[12]  Amir F. Atiya,et al.  An efficient stock market forecasting model using neural networks , 1997, Proceedings of International Conference on Neural Networks (ICNN'97).

[13]  Tomaso A. Poggio,et al.  Machine Learning, Machine Vision, and the Brain , 1999, AI Mag..

[14]  J. Shadbolt,et al.  Neural network system for tactical asset allocation in the global bonds markets , 1993 .

[15]  Mary Malliaris.  Modeling the behavior of the S&P 500 index: a neural network approach , 1994, Proceedings of the Tenth Conference on Artificial Intelligence for Applications.

[16]  Paul J. Werbos.  Backpropagation: basics and new developments , 1998 .

[17]  Y. Hiemstra.  Applying Neural Networks and Genetic Algorithms to Tactical Asset Allocation , 1996 .