Advances and Applications in Statistics
Volume 49, Issue 4, Pages 287 - 303
(October 2016) http://dx.doi.org/10.17654/AS049040287 |
|
PARAMETER ESTIMATION OF MULTIVARIATE SCALED T DISTRIBUTION: AN APPLICATION TO ALL SHARE PRICE INDEX
N. V. Chandrasekara, M. A. Mammadov and C. D. Tilakaratne
|
Abstract: Identifying multivariate distributions related to practical scenarios became an interesting research area among scholars recently as most of the practical scenarios are associated with more than one variable. Analytical methods have been introduced by many investigators to estimate parameters of multivariate distributions. However, they involve heavy mathematical calculations when the number of variables associated with the problem increases. Numerical methods were used by many researchers to find parameters of multivariate distributions and many were limited to combination of few variables. The objective of this study is to estimate the parameters of the multivariate scaled t distribution of the return series of all share price index (ASPI) of Colombo stock exchange and those of its associated financial indices. Four financial indices: Amex Oil Index, Amex Gold Index, World Cocoa Index and the US GSPC index of which are associated with ASPI were considered in the study. Corresponding optimization problem involves twenty one parameters related to five dimensions of the multivariate scaled t distribution. A local optimization method and a global optimization method were used to solve the problem. Results exhibit that the global optimization method provides substantially better optimal values for the parameters of the multivariate scaled t distribution. |
Keywords and phrases: multivariate distribution, scaled t distribution, all share price index, local optimization, global optimization. |
|
Number of Downloads: 410 | Number of Views: 1474 |
|