ABSTRACT
Adaptive filtering is a topic of enormous practical relevance and covers theoretical challenges that persist even today. Techniques related to adaptive filtering have been widely used in numerous prediction and forecasting tasks. The main purpose of this chapter is to present the impact of the most state-of-the-art adaptive filtering techniques (least mean squares, recursive least squares, incremental delta-bar-delta and Kalman filter) in the forecasting problem of some renowned derivative indices. The indices examined on forecasting their upcoming trends are three noisy derivative financial signals: CAC40, DAX30 and the Euronext. In order to test the adaptive filtering techniques’ forecasting effectiveness, we use four traditional strategies (naive strategy, buy and hold strategy, MACD and a simple ARMA) as a benchmark reference. The performance of the adaptive filtering techniques proved to greatly overcome the performance of the traditional strategies. Further discussion on the underlying mechanisms that led to the better performance of the adaptive filtering techniques is carried out.
