Stock Price Prediction using Transformer Models

Author

Manoj Chandrasekharan

Date

2024

Document Type

Thesis (Access Restricted)

Degree Name

Master of Science

Department

Electrical & Computer Engineering

Committee Chair

Bonny Banerjee

Committee Member

Madhusudhanan Balasubramanian

Committee Member

Pankaj Jain

Abstract

Stock price prediction is an important and widely studied problem. We use two variants of a multimodal transformer model to predict future stock price from past stock price and trading volume. We experimented on predicting the S&P 500, DJI, NASDAQ, and the benchmark StockNet dataset. Our evaluations using standard metrics and comparison to the state-of-the-art reveal that the multimodal transformer yields higher accuracy in some cases but falls short in some others. We also report the accuracy of the models for multiple time horizons, multiple input durations, two kinds of data normalization techniques, and different model sizes.

Comments

Data is provided by the student.”

Library Comment

Dissertation or thesis originally submitted to ProQuest.

Notes

No Access

Recommended Citation

Chandrasekharan, Manoj, "Stock Price Prediction using Transformer Models" (2024). Electronic Theses and Dissertations. 3688.
https://digitalcommons.memphis.edu/etd/3688