Work detail

Practical Use of Neural Networks on the Financial Markets

Author: Mgr. Tomáš Slabý
Year: 2002 - summer
Leaders:
Consultants:
Work type: Financial Markets
Masters
Language: Czech
Pages: 80
Awards and prizes:
Link:
Abstract: The goal of this thesis is a qualitative comparison of neural networks to the classical statistic methods on the financial markets. I focus mainly on the practical use, because so we can gain further important information and experiences about the instrument.
Neural networks are a subset of artificial intelligence systems. This is a kind of method, which in a way seems like a doing of an intelligent subject or like a process that takes place in nature. We can use this methodology as an alternative to e. g. some linear analysis on the financial markets.
After I focus on a general overview of artificial intelligence systems and the technique of some neural networks, I discuss three types of problems: classification and regression problem and time series analysis. I have compared advantages and disadvantages of the use of such neural networks to discrimination analysis (capture no. 5), to linear regression (capture no. 6) and to the classical time series analysis (capture no. 8), such as the ARIMA, exponencial smoothing and various types and modifications of moving averages in expert systems. I have tested the efficient market hypothesis on the index PX-50 in the capture no. 7.



Partners

Deloitte

Sponsors

CRIF
McKinsey
Patria Finance