This paper introduces a Bidirectional Multivariate LSTM model for short-term stock price forecasting in the Indian market. Using eight years of hourly data from four major NIFTY 100 companies, the model incorporates OHCLV metrics and 12 highly correlated technical indicators. Comparative analysis shows the proposed model outperforms traditional LSTM approaches in predictive accuracy. The model achieves an average R2 score of 99.48% and low error rates across multiple metrics. These results demonstrate its strong potential for enhancing short-term trading strategies.
Omkar Oak$^{*}$ , Rukmini Nazre, Rujuta Budke and Yogita Mahatekar
Our research evaluates different approaches to stock price prediction:
In the univariate approach, we use only the closing price from the dataset to predict future stock prices. This simple method serves as our baseline for comparison.
The multivariate OHCLV approach incorporates Open, High, Close, Low, and Volume data to provide more context for the prediction model, allowing it to capture more complex relationships in the market data.
Our proposed Bidirectional LSTM approach incorporates the following technical indicators which have high correlation values (>0.99) with the close price, in addition to OHCLV metrics.
If you use this code or the findings in your research, please cite our paper:
@INPROCEEDINGS{10923989,
author={Oak, Omkar and Nazre, Rukmini and Budke, Rujuta and Mahatekar, Yogita},
booktitle={2024 5th IEEE Global Conference for Advancement in Technology (GCAT)},
title={A Novel Multivariate Bi-LSTM model for Short-Term Equity Price Forecasting},
year={2024},
pages={1-6},
doi={10.1109/GCAT62922.2024.10923989}}
For any questions or inquiries about this research, please open an issue on this repository or contact the corresponding author.