Artificial Neural Networks and ARIMA-Models within the Field of Stock Market Prediction - A Comparison

摘要: Abstract Within the field of stock market prediction a controversial discussion between technicians and fundamentalists concerning qualification these different methods has taken place. On one hand, experts use so-called charts to extract those formations they regard be significant for future development prices. This procedure requires extensive experience in recognizing interpreting patterns can also contain many sources error. other have decide which information, even regarding influences, consider. Therefore, it is intended link both perspectives. Some analysts statistical (i.e. moving averages or auto-regressive models) order indicate important clues trends The ARIMA-Model combines abilities two methods. Another problem-solving approach uses Artificial Neural Networks (ANN). They are loose sense based on concepts derived from research into nature brain [16]. Particularly ANN's ability filtering ‚noisy’ may caused by differential behaviour various investors seems predetermine this approach. Our intention approaches short-term (the following day's price). In spite that will extended medium-term (a monthly forecast).