ELLIOTT WAVES RECOGNITION VIA NEURAL NETWORKS Martin Kotyrba Eva Volna David Brazina Robert Jarusek Department of Informatics and Computers University of Ostrava Z70103, Ostrava, Czech Republic martin.kotyrba@osu.cz eva.volna@osu.cz david.brazina@osu.cz robert.jarusek@osu.cz KEYWORDS Elliott wave, neural networks, pattern recognition. ABSTRACT In this paper we introduce our method that is able to analyze and recognize Elliott waves in time series. Our method uses an artificial neural network that is adapted by backpropagation. Neural network uses Elliot wave’s patterns in order to extract them and recognize. Artificial neural networks are suitable for pattern recognition in time series mainly because of learning only from examples. There is no need to add additional information that could bring more confusion than recognition effect. Neural networks are able to generalize and are resistant to noise. On the other hand, it is generally not possible to determine exactly what a neural network learned and it is also hard to estimate possible recognition error. They are ideal especially when we do not have any other description of the observed series. This paper also includes experimental results of Elliott waves recognition carried out with our method. INTRODUCTION Financial data is a set of economic indexes with certain significance in finance. The basic pattern recognition can be classified according to various criteria. Basic criteria of time series pattern recognition are to identify the direction of the trend. The basic step for the future prediction is patterns recognition. The Elliott Wave Principle is a detailed description of how groups of people behave. It reveals that mass psychology swings from pessimism to optimism and back in a natural sequence, creating specific and measurable patterns. One of the easiest places to see the Elliott Wave Principle at work is in the financial markets, where changing investor psychology is recorded in the form of price movements. If we can identify repeating patterns in prices, and figure out where we are in those repeating patterns today, we can predict future trend. This paper introduces our method that allows analysis Elliot wave’s patterns in time series for the purpose of their future prediction. ARTIFICIAL NEURAL NETWORKS An Artificial Neural Network (ANN) is a connectionist massively parallel system, inspired by the human neural system. Its units, neurons (Fig. 1), are interconnected by connections called synapse. Each neuron, as the main computational unit, performs only a very simple operation: it sums its weighted inputs and applies a certain activation function on the sum. Such a value then represents the output of the neuron. However great such a simplification is (according to the biological neuron), it has been found as plausible enough and is successfully used in many types of ANN, (Fausett 1994). Figures 1: Model of neuron A neuron X i obtains input signals x i and relevant weights of connections w i , optionally a value called bias b i is added in order to shift the sum relative to the origin. The weighted sum of inputs is computed and the bias is added so that we obtain a value called stimulus or inner potential of the neuron s i . After that it is transformed by an activation function f into output value o i that is computed as it is shown in equations (see Fig.1) and may be propagated to other neurons as their input or be considered as an output of the network. Here, the activation function is a sigmoid, (Kondratenko and Kuperin 2003). The purpose of the activation function is to perform a threshold operation on the potential of the neuron. Proceedings 26th European Conference on Modelling and Simulation ©ECMS Klaus G. Troitzsch, Michael Möhring, Ulf Lotzmann (Editors) ISBN: 978-0-9564944-4-3 / ISBN: 978-0-9564944-5-0 (CD)