Models of capital market investors' behavior

This paper aims to study the behavior described by the theory of capital markets from the ones suggested by the efficient market theory up to chaotic behavior and fractal market theory, investigating the possibility of forecasting financial time series behavior through artificial intelligence concepts and tools (artificial neural networks, fuzzy logic, neuro-fuzzy hybrid systems).

 

The first chapter of the book emphasizes the study of classical models described through the efficient market behavior, following the lead in the attempt to forecast financial time series for modern portfolio theory, Capital Asset Pricing Model, Arbitrage Pricing Theory, Fama-French model and other traditional techniques.

In the second chapter is examined the scientific and research framework in which was developed the chaos theory, multiple definitions, methods of analysis and measurement, emphasizing investigation on available tests for detecting the presence of nonlinearity and chaotic behavior on the capital market in Romania. In this part of the book are presented different methodologies of the tests used for detecting chaos (Lyapunov exponents, correlation dimension, BDS statistics, Hinich bispectral test, NEGM test, White test, Kaplan test).

On the third chapter proposes the approach and the interpretation of economic phenomena and characterizing the stock market behavior in terms of fractal geometry. This distinct view to capital markets offers a different connotation to the road between chaos and order and is based on fractal market hypothesis. The particular properties and methods of the fractal analysis are conceptually investigated through R/S analysis, fractal dimension analysis, fractal distribution analysis, DFA and MF-DFA models.

In the structure of the last chapter of the book are presented the concepts, the architecture, the components and the relationships between them, defining artificial neural networks, fuzzy-logic systems and hybrid neuro-fuzzy networks. The study accomplished in this part of the book aims to compare the classical methods of forecasting with those based on artificial intelligence, in the attempt to predict the prices in the Romanian stock market.

 

The usefulness of the research conducted in this paper consists in the possibility of practical use of various models, methods and techniques presented and implemented as tools for predicting the evolution of financial time series, designed for financial intermediaries, offering a wide range of approaches that describe different behaviors (efficient capital market, deterministic nonlinear behavior, fractal behavior, artificial neural networks, fuzzy and neuro-fuzzy logic systems) of the capital market. From the conceptual point of view, the usefulness of this research derives from the multiple theories investigation, from classical to the latest interdisciplinary approaches.

The book is recommended for students, researchers and all those interested in the capital markets.