Journal article Open Access
New Log Likelihood Estimation Function
Louangrath, P.
This paper provides a New Log-Likelihood Estimator (NLLE) function as a tool for value approximation. We improved the accuracy of the log MLE in two steps (i) determine the log likelihood of a random variable X, and (ii) adjust the estimate by a factor of . In-Sample testing was accomplished by using daily SET100 indices over a period of 60 days. Out-of-sample data were used for confirmatory verification; out-of-sample data came from 5 major stock markets: NASDAQ, DOW, SP500, DAX, and CAC40. Relevant tests used to compare the results of the proposed NLLE include Cramer-Rao Lower Bound (CRLB), Likelihood Ratio Test, Wald statistic, and Lagrange Multiplier (Score Statistic). It was found that NLLE is more efficient than the conventional MLE. It gives practitioners a better tool for value estimation in many fields of natural and social sciences.
| Name | Size | |
|---|---|---|
|
ARTICLE 3, Vol 1, No 2, New Log Likelihood Estimation Function.pdf
md5:60023cfba70d43dd319df80cd89a6465 |
133.6 kB | Download |
| All versions | This version | |
|---|---|---|
| Views | 23 | 19 |
| Downloads | 31 | 25 |
| Data volume | 3.8 MB | 3.3 MB |
| Unique views | 20 | 18 |
| Unique downloads | 25 | 23 |
- Publication date:
- June 30, 2015
- DOI:
-
- Keyword(s):
- data types, quantitative data, nominal data, ordinal data
- Published in:
- Inter. J. Res. Methodol. Soc. Sci: 1 pp. 36-47 (2).
- License (for files):
- Creative Commons Attribution 4.0 International