Data sets and code for "Outlier detection of multivariate data via the maximization of the cumulant generating function"

Anonymized data sets and MATLAB code for the paper "Outlier detection of multivariate data via the maximization of the cumulant generating function" (F. Cesarone, R. Giacometti, J.M. Ricci), Journal of Computational and Applied Mathematics, Vol. 461, 116457 (2025), https://doi.org/10.1016/j.cam.2024.116457. The package contains: (i) the daily linear returns (computed from prices adjusted for splits and dividends) of the constituents of six markets, in .xlsx and .mat format, from 17/02/1997 to 14/02/2022 (6521 daily observations): Dow Jones Industrial Average (30 assets), Euro Stoxx 50 (47 assets), DAX 30 (28 assets), CAC 40 (40 assets), FTSE 100 (98 assets) and EuroNext 100 (98 assets); (ii) the MATLAB functions ProjGrad.m (maximization of the cumulant generating function, Table 1 of the paper) and CumulantsOutDet.m (outlier detection, Table 2 of the paper). Because of file size limits, the data are split into one zip file per market (for FTSE 100 and EuroNext 100, the .mat and .xlsx files are provided separately). The data are anonymized: asset names, tickers and dates have been removed, and each file contains only numeric matrices (rows are time periods in chronological order, columns are assets).