A NEURAL NETWORK MODEL FOR OPTION PRICING
LI YILI
LI YILI
Citations
Altmetric:
Alternative Title
Abstract
This study concentrates on the applications of neural networks on option pricing
The traditional Black-Scholes model systematically misprices options due to the incorporation of unrealistic assumptions. On the other hand, neural networks , which allow the data to determine the structure and parameters of a system, provide a very general framework for various pricing functions. Therefore, a neural network model may be a good alternative to the traditional Black-Scholes model. In this study, a multi-layer feed-forward Back-propagation neural network is applied in the construction of an option pricing model. Our neural network model does not incorporate the many unrealistic assumptions required by the Black-Scholes model.
Based on this model, a neural network based option pricing system is developed.
Tests carried out to price Nikkei 225 stock index call option prices using this system
yield more accurate pricing than results obtained using the traditional Black-Scholes model. Specifically, the data sample is categorized into three parts: deep in-the-money, at-the-money and deep out-of-the-money. For each category, the Black-Scholes model is applied and the results obtained shows significant deviations between the actual market prices and the Black-Scholes model prices for deep in-the-money and out-of-the-money contracts. Likewise, our neural network model is applied to each category of the data sample. More accurate results are obtained, especially for deep in-the-money and out-of-the-money contracts which are mispriced by Black-Scholes model . Hence, neural networks based models are a good alternative to the traditional Black-Scholes model.
Keywords
Source Title
Publisher
Series/Report No.
Collections
Rights
Date
1997
DOI
Type
Thesis