Output details
11 - Computer Science and Informatics
University of York
A learning adaptive Bollinger band system
<24>CiFEr is the main IEEE conference in its area. This article contributes to the state-of-the-art in financial forecasting through the development of a novel adaptive approach based on heterogeneous meta-learning, the strengths of which are confirmed on both simulated and real-world data. A methodological contribution to machine learning is also made as the Learning Classifier System paradigm is extended to make use of a population of signal processors other than classifiers. The resulting tool extends substantially the potential of a popular financial indicator (Bollinger Bands) by optimising its parameters, and exploiting the power of a population of such indicators.