Open AccessMultiple kernel support vector regression for pricing nifty option
Author(s) -
Neetu Verma,
Sujoy Das,
Namita Srivastava
Publication year - 2015
Publication title -
international journal of applied mathematical research
Language(s) - English
Resource type - Journals
ISSN - 2227-4324
DOI - 10.14419/ijamr.v4i4.5023
Subject(s) - support vector machine , econometrics , mathematics , kernel (algebra) , regression , mean squared error , context (archaeology) , regression analysis , kernel regression , valuation of options , statistics , computer science , artificial intelligence , paleontology , combinatorics , biology
The goal of present experiments is to investigate the use of multiple kernel learning as a tool for pricing options in the context of Indian stock market for Nifty index options. In this paper, fair price of an option is predicted by Multiple Kernel Support Vector Regression (MKLSVR) using linear combinations of kernels and Single Kernel Support Vector Regression (SKSVR). Prices of option highly depend on different money market conditions like deep-in-the-money, in-the-money, at-the-money, out-of-money and deep-out-of-money condition. The experimental study attempts to identify the forecasting errors with the help of mean square error; root meant square error, and normalized root meant square error between the market option prices and the calculated option prices by model for all market conditions. The results reflect that multiple kernel support vector regression performed fairly well in comparison to support vector regression with single kernel.