Open AccessClustering and Classification in Option Pricing
Author(s) -
Nikola Gradojević,
Dragan Kukolj,
Ramazan Gençay
Publication year - 2011
Publication title -
review of economic analysis
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.101
H-Index - 1
ISSN - 1973-3909
DOI - 10.15353/rea.v3i2.1458
Subject(s) - cluster analysis , valuation of options , generalization , computer science , artificial neural network , modular design , vector quantization , artificial intelligence , machine learning , economics , econometrics , mathematics , mathematical analysis , operating system
This paper reviews the recent option pricing literature and investigates how clustering and classification can assist option pricing models. Specifically, we consider non-parametric modular neural network (MNN) models to price the S&P-500 European call options. The focus is on decomposing and classifying options data into a number of sub-models across moneyness and maturity ranges that are processed individually. The fuzzy learning vector quantization (FLVQ) algorithm we propose generates decision regions (i.e., option classes) divided by ‘intelligent’ classification boundaries. Such an approach improves generalization properties of the MNN model and thereby increases its pricing accuracy.