A method of parameterising a feed forward multi-layered perceptron artificial neural network, with reference to South African financial markets

摘要

No analytic procedures currently exist for determining optimal artificial neural network structures andparameters for any given application. Traditionally, when artificial neural networks have been appliedto financial modelling problems, structure and parameter choices are often made a priori withoutsufficient consideration of the effect of such choices. A key aim of this study is to develop a generalmethod that could be used to construct artificial neural networks by exploring the model structure andparameter space so that informed decisions could be made relating to the model design. In this study,a formal approach is followed to determine suitable structures and parameters for a Feed ForwardMulti-layered Perceptron artificial neural network with a Resilient Propagation learning algorithm witha single hidden layer. This approach is demonstrated through the modelling of four South Africaneconomic variables, namely the average monthly returns on the money, bond and equity markets aswell as monthly inflation. Artificial neural networks can be constructed on the aforementioned variablesin isolation or, jointly, in an integrated model. The performance of a range of more traditional timeseries models is compared with that of the artificial neural network models. The results suggest that,on a statistical level, artificial neural networks perform as well as time series models at forecasting thereturns for financial markets. Hybrid models, combining artificial neural networks with the time seriesmodels, are constructed, trained and tested for the money market and for the rate of inflation. Theyappear to add value to the time series models when forecasting inflation, but not for the money market.