Empirical performance of interpolation techniques in risk-neutral density (RND) estimation

37th International Conference on Quantum Probability and Related Topics

Bahaludin, H.^1 ; Abdullah, M.H.^1

Department of Computational and Theoretical Sciences, Kulliyyah of Science, International Islamic University Malaysia, Bandar Indera Mahkota Campus, Jalan Sultan Ahmad Shah, Pahang, Kuantan

25200, Malaysia^1

关键词: Empirical performance;    Fourth order polynomial;    Interpolation techniques;    Leave-one-out cross-validation (LOOCV);    Risk neutrals;    Second-order polynomial;    Smoothing spline;    Statistical differences;   

Others  :  https://iopscience.iop.org/article/10.1088/1742-6596/819/1/012026/pdf
DOI  :  10.1088/1742-6596/819/1/012026

来源: IOP

【 摘 要 】

The objective of this study is to evaluate the empirical performance of interpolation techniques in risk-neutral density (RND) estimation. Firstly, the empirical performance is evaluated by using statistical analysis based on the implied mean and the implied variance of RND. Secondly, the interpolation performance is measured based on pricing error. We propose using the leave-one-out cross-validation (LOOCV) pricing error for interpolation selection purposes. The statistical analyses indicate that there are statistical differences between the interpolation techniques:second-order polynomial, fourth-order polynomial and smoothing spline. The results of LOOCV pricing error shows that interpolation by using fourth-order polynomial provides the best fitting to option prices in which it has the lowest value error.