A nonparametric outlier detection method for financial data

Outlier detection has many important applications in financial surveillance. Mining outliers in database is to find exceptional objects that deviate from the rest of the data set. The Traditional outlier detection method is based on statistical models, such as ARMA and ARCH, which require special hypotheses and try to describe the system behavior by a fixed structure. The statistical models are inappropriate to apply to complex financial data, such as high-frequency data. This paper introduces a nonparametric method to detect outliers for financial data. Based on the Voronoi diagram, we propose a novel outlier detection method, which called Voronoi based Outlier Detection (VOD). Experiments show the VOD method performs effective in outlier detection for both daily and ultra-high-frequency financial data.