Time Series Prediction Using Complex Methods

OData support
Supervisor:
Dr. Vajda Ferenc
Department of Control Engineering and Information Technology

The present study aims to examine time series prediction in different perspective than it was done previously by other methods. The main purpose of fuzzy mathematics is better modelling social, cognitive and environmental processes, in which cases the uncertainity must be managed more effectively than before. With fuzzy logic it is possible to use the tools of technical analysis to predict different time series. The success of technical analysis depends on how one interprets the available signals. The global behaviour of the system can be described with linguistic rules, so integration of this interpretation and human expertise makes effective financial time series prediction implicitly available. The realized system examines the adaptive approach in a new perspective so far beyond the former proposals. The resulting configurable trading robot can be customized to any currancy pair. The system will be tested in various currency pairs. The emprical results show that the proposed system is capable of generating higher risk-discounted returns, and in many cases it overperforms the Buy and Hold strategy. The study briefly deals with the opportunities for the further development: paramteroptimalization and creation of an integrated system.

Downloads

Please sign in to download the files of this thesis.