NEURAL NETWORKS : AN APPLICATION OF BACKPROPAGATION TO SINGAPORE WARRANT PRICING

Abstract

The purpose of this thesis is to investigate the backpropagation neural network algorithm as put forward by Wcrbos (1974) and then refined by Rumelhart, Hinton and Williams (1986). In the context of the Stock Exchange of Singapore (SES), it wishes to examine the efficiency of a warrant pricing neural network created by this algorithm. It also wishes to see if it is possible to use such a network to trade warrants and obtain abnormal returns. The layout of this thesis will be as follows:

Chapter 1 A brief literature review of neural networks. This includes a brief history of neural networks and an introduction to the basics of how such networks work. There will also be a simple example of how a neural network is able to identify an XOR function.

Chapter 2 Backpropagation. This chapter introduces the neural network algorithm that is the focus of this thesis. It gives a brief derivation of the basic learning law and discusses modifications proposed by other writers in the field.

Chapter 3 An introduction to the basic Black-Scholes European Call Pricing model. Following this, modifications to this model so that it can be used to price Singapore warrants arc discussed. The reason for doing this is to present an alternative model against which the backpropagation created neural network model can be compared.

Chapter 4 The empirical results from the estimation of the neural network models and Black-Scholes models arc presented. Based on these findings, this thesis will draw a conclusion as to which is a better model to use.

Chapter 5 Based on the created neural network model, it is investigated if it is possible to use it in conjunction with trading rules to generate abnonnal returns. Here the Capital Asset Pricing Model (CAPM) is used to judge if returns are theoretically abnormal. A passive buy and hold market strategy is used to see if the neural network trading system could outperfonn the market under realistic trading conditions.