Expectation maximization algorithm for multivariate exponential power mixture model with applications to financial data
| dc.contributor.author | Ojemola, Oladipupo Ibukun | |
| dc.date.accessioned | 2023-05-13T16:55:55Z | |
| dc.date.available | 2023-05-13T16:55:55Z | |
| dc.date.issued | 2015 | |
| dc.description | vii,64p | en_US |
| dc.description.sponsorship | This study obtained the parameter estimates of the multivariate exponential power distribution and fitted the model on stock returns data of three stocks on the Nigerian Stock Exchange market and compared the Value at Risk forecasts when multivariate exponential power mixture distribution was the underlying distribution of the data with the multivariate normal mixture Value at Risk. This was with a view to reducing possible loss on financial assets. The parameters of the multivariate normal mixture and multivariate exponential power mixture model were obtained by Expectation Maximization algorithm. The method of probability fitting was used to fit three company shares quoted on the Nigerian Stock Exchange market and the best fit model was determined using Pearson Chi-Square approach. Value-at-Risk (VaR) forecasts were obtained using the fitted model parameter estimates of the multivariate power mixture model and compared with the multivariate normal mixture model. After 25 iterations, the parameter estimates and the log-likelihood of the multivariate exponential power mixture model were obtained. Also, at both 95% and 99% confidence levels, it was discovered that the estimates for the VaR were lower when the multivariate mixture normal model was assumed than when the multivariate exponential power mixture model was assumed. The study concluded that Value at Risk forecast of a portfolio of asset returns would be more reliable if exponential power distribution was modelled on such data. | en_US |
| dc.identifier.citation | Ojemola,O.I(2015).Expectation maximization algorithm for multivariate exponential power mixture model with applications to financial data. Obafemi Awolowo University. | en_US |
| dc.identifier.uri | https://ir.oauife.edu.ng/123456789/5407 | |
| dc.language.iso | en | en_US |
| dc.publisher | Mathematics,Obafemi Awolowo University. | en_US |
| dc.subject | Exponential power | en_US |
| dc.subject | Multivariate exponential | en_US |
| dc.subject | Stock exchange | en_US |
| dc.subject | Algorithm | en_US |
| dc.title | Expectation maximization algorithm for multivariate exponential power mixture model with applications to financial data | en_US |
| dc.type | Thesis | en_US |