Abstract
A novel hybrid option pricing model using a deep learning neural network has been developed. The hybrid model keeps the traditional option pricing model with the same input parameters while simultaneously adjusting the model with neural network methods to improve accuracy when applied to real market data, especially in OTM options. The new hybrid model demonstrates superior accuracy compared to both traditional parametric and non-parametric option pricing models for both Call and Put options across all moneyness levels. The empirical results of the hybrid model provide an explanation for the deviation from the Put-Call parity observed in real market data.
Original language | English |
---|---|
Article number | 123979 |
Journal | Expert Systems with Applications |
Volume | 251 |
DOIs | |
State | Published - 1 Oct 2024 |
Externally published | Yes |
Bibliographical note
Publisher Copyright:© 2024 Elsevier Ltd
Keywords
- Computational finance
- Neural networks
- Option pricing
- Put-call parity