The moving window versus the multiresolution algorithm: A filtering outlier detection comparison

• Autores: Josep Fortiana Gregori, Michèle Murez, Josep María Puigvert Gutiérrez
• Localización: XXXI Congreso Nacional de Estadística e Investigación Operativa ; V Jornadas de Estadística Pública: Murcia, 10-13 de febrero de 2009 : Libro de Actas, 2009, ISBN 978-84-691-8159-1
• Idioma: inglés
• Texto completo no disponible (Saber más ...)
• Resumen

  • Financial market data might contain outliers or suspicious observations which may be dicult to detect using informal inspection and graphical displays. Puigvert and Fortiana (2008) propose and study the performance of a ltering procedure based on a moving window algorithm. In this study we compare two techniques in order to detect outliers in nancial market data. The moving window ltering algorithm is tested against a second technique which is based on wavelet analysis. We apply both algorithms to a set of nancial market data which consists of 25 series selected from a larger dataset using a cluster analysis technique taking into account the daily behaviour of the market; each of these series is a representative of a cluster that represents a di erent segment of the market. Although both algorithms seem to detect most of the outliers introduced, there are still some caveats that should be taken into account when using the outlier detection algorithm depending on the cluster.