Two stages fitting techniques using generalized lambda distribution: application on Malaysian financial return

Abstract

The underline distribution assumption used in the analysis of share market returns is crucial in risk management. An important aspect related to stock return modelling is to obtain accurate prediction. This paper presents an innovative fitting method called two stages (TS) method for modelling daily stock returns. The proposed approach by first establishing trend in the series, and then separately performing L-moment estimation on the generalized lambda distribution (GLD) parameter. The performance of the TS-GLD models had been evaluated using Monte Carlo simulation and Malaysian Kuala Lumpur Composite Index (KLCI) returns from year 2001 to 2015. Based on k-sample Anderson darling goodness of fit test, the two stages GLD model in location parameter (GLD.1) performed well in all studied cases. The GLD.1 model benefits risk management by providing effective distribution fitting.