American Options Pricing Using Neural Networks

Introduction

In this repository, we explore the application of various neural network models to accurately predict American call options prices. American options, known for their flexible exercise terms, present a unique challenge in financial modeling. This project aims to bridge the gap in conventional models with advanced neural network techniques, offering a more nuanced understanding of these complex financial instruments.

About the Research Paper

The accompanying research paper delves into the use of neural networks in financial markets, specifically for predicting American call option prices. It compares several neural network models, including Feedforward Neural Networks (FNNs), Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and Modular Neural Networks (MNNs), against traditional methods like the Barone-Adesi and Whaley (B-AW) approximation. The paper can be found in the repository as American Options Pricing Using Neural Networks.pdf.

Repository Structure

American Options Pricing Using Neural Networks.pdf - The research paper detailing the study and its findings.

Barone-Adesi and Whaley Approximation and FNN.ipynb - Jupyter notebook implementing the Barone-Adesi and Whaley model and implementation of the Feedforward Neural Networks.

Data Exploration.ipynb - Notebook for initial data analysis and exploration.

Data Preprocessing.ipynb - Details the data preprocessing steps used in the models.

LSTM + CNN.ipynb - Implementation combining Long Short-Term Memory networks and CNNs.

MNN.ipynb - Modular Neural Networks application in American options pricing.

options_data - Folder containing options data txt files of stocks: AAPL, NVDA and TSLA from Jan 2023 to Sept 2023

CONTRIBUTORS.md - A file listing the contributors to this project.

How to Use

Clone the repository to get a local copy. Ensure you have the necessary dependencies installed, as outlined in each Jupyter notebook. Run Data Preprocessing.ipynb first to generate the data.csv. Please make sure to get your own FRED API KEY. Run the notebooks to understand the implementation details and experiment with the models.