TY - GEN
T1 - L q-regularization of the Kalman Filter for exogenous outlier removal
T2 - 2011 4th IEEE International Workshop on Computational Advances in Multi-Sensor Adaptive Processing, CAMSAP 2011
AU - Jay, Emmanuelle
AU - Duvaut, Patrick
AU - Darolles, Serge
AU - Gourieroux, Christian
PY - 2011/12/1
Y1 - 2011/12/1
N2 - This paper presents a simple and efficient exogenous outlier detection & estimation algorithm introduced in a regularized version of the Kalman Filter (KF). Exogenous outliers that may occur in the observations are considered as an additional stochastic impulse process in the KF observation equation that requires a regularization of the innovation in the KF recursive equations. Regularizing with a l 1- or l 2-norm needs to determine the value of the regularization parameter. Since the KF innovation error is assumed to be Gaussian we propose to first detect the possible occurrence of an exogenous impulsive spike and then to estimate its amplitude using an adapted value of the regularization parameter. The algorithm is first validated on synthetic data and then applied to a concrete financial case that deals with the analysis of hedge fund returns. The proposed algorithm can detect anomalies frequently observed in hedge returns such as illiquidity issues.
AB - This paper presents a simple and efficient exogenous outlier detection & estimation algorithm introduced in a regularized version of the Kalman Filter (KF). Exogenous outliers that may occur in the observations are considered as an additional stochastic impulse process in the KF observation equation that requires a regularization of the innovation in the KF recursive equations. Regularizing with a l 1- or l 2-norm needs to determine the value of the regularization parameter. Since the KF innovation error is assumed to be Gaussian we propose to first detect the possible occurrence of an exogenous impulsive spike and then to estimate its amplitude using an adapted value of the regularization parameter. The algorithm is first validated on synthetic data and then applied to a concrete financial case that deals with the analysis of hedge fund returns. The proposed algorithm can detect anomalies frequently observed in hedge returns such as illiquidity issues.
UR - https://www.scopus.com/pages/publications/84857149043
U2 - 10.1109/CAMSAP.2011.6136009
DO - 10.1109/CAMSAP.2011.6136009
M3 - Conference contribution
AN - SCOPUS:84857149043
SN - 9781457721052
T3 - 2011 4th IEEE International Workshop on Computational Advances in Multi-Sensor Adaptive Processing, CAMSAP 2011
SP - 29
EP - 32
BT - 2011 4th IEEE International Workshop on Computational Advances in Multi-Sensor Adaptive Processing, CAMSAP 2011
Y2 - 13 December 2011 through 16 December 2011
ER -