Portfolio Construction Based on LSTM RNN and Black-Litterman Model: Evidence from Yahoo Finance

Junzhe Wang

2024

Abstract

Portfolio optimization is always a tough issue in fiance field. This study explores the integration of Long Short-Term Memory (LSTM) Recurrent Neural Networks (RNN) with the Black-Litterman model (BL model) to improve portfolio optimization. The BL model, which combines the views from the investors with market equilibrium to modify the revenues that are expected, is commonly used for asset allocation. Yet the model has a few limitations, including subjectivity, data sensitivity, and complexity. In order to deal with these issues, the paper proposes incorporating LSTM RNN predictions into the BL model to mitigate bias and enhance decision-making. The study utilizes historical data from Yahoo Finance for four major corporations(Apple;Intel;Google;NVIDIA) from January 2023 to August 2024. The LSTM RNN is trained on this data to generate machine predictions, which are then treated as investor views in the BL model. The Omega matrix, representing the uncertainty or confidence in these predictions, is adjusted to combine machine and investor perspectives. Results indicate that while LSTM RNN predictions can improve price forecasting, they also introduce biases that require careful calibration. The modified BL model, incorporating machine-generated views, provides a more personalized and potentially more accurate portfolio allocation. This approach offers a novel way to balance human and machine insights in financial decision-making, though it requires significant computational resources and expertise to implement effectively. Future research might focus on refining the Omega matrix estimation and exploring alternative machine learning models to further enhance the model's robustness.

Paper Citation

in Harvard Style

Wang J. (2024). Portfolio Construction Based on LSTM RNN and Black-Litterman Model: Evidence from Yahoo Finance. In Proceedings of the 1st International Conference on E-commerce and Artificial Intelligence - Volume 1: ECAI; ISBN 978-989-758-726-9, SciTePress, pages 337-343. DOI: 10.5220/0013225400004568

in Bibtex Style

@conference{ecai24,
author={Junzhe Wang},
title={Portfolio Construction Based on LSTM RNN and Black-Litterman Model: Evidence from Yahoo Finance},
booktitle={Proceedings of the 1st International Conference on E-commerce and Artificial Intelligence - Volume 1: ECAI},
year={2024},
pages={337-343},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013225400004568},
isbn={978-989-758-726-9},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on E-commerce and Artificial Intelligence - Volume 1: ECAI
TI - Portfolio Construction Based on LSTM RNN and Black-Litterman Model: Evidence from Yahoo Finance
SN - 978-989-758-726-9
AU - Wang J.
PY - 2024
SP - 337
EP - 343
DO - 10.5220/0013225400004568
PB - SciTePress