Stock price prediction using neural networks: A case of Pakistan
Abstract/Description
This study investigated the performance of two popular neural network architectures, LSTM and GRU, in predicting stock prices. Five-year daily price data of four major Pakistan public companies based on the highest market capital in their respective sectors were used for the analysis. The data was split into training and testing data. The models were also tested using SVM and DT models as performance comparisons with the neural networks. The results of this study showed that the LSTM and GRU models are the most successful in terms of average returns when predicting stock prices because of the ability to capture long-term dependencies in the data, which is beneficial for predicting future stock prices. The SVM and DT models, on the other hand, are much less successful due to their inability to capture long-term dependencies. Based on the findings of this study, it is recommended that investors and traders use the LSTM and GRU models when predicting stock prices. Furthermore, it is recommended that further research be conducted to explore the potential of other models, such as deep learning models, for predicting stock prices.
Keywords
Neural Network Architectures, LSTM, GRU, Stock Prices, Long-Term Dependencies, Performance Comparisons
Track
Finance
Session Number/Theme
Session 1B: Finance
Session Chair
Dr. Sana Tauseef ; Dr. Mohsin Khawaja
Start Date/Time
26-5-2023 2:45 PM
End Date/Time
26-5-2023 4:45 PM
Location
MCS-4, Aman-CED, First Floor, Institute of Business Administration, Karachi
Recommended Citation
., M. (2023). Stock price prediction using neural networks: A case of Pakistan. 3rd IBA SBS International Conference 2024. Retrieved from https://ir.iba.edu.pk/sbsic/2023/program/7
COinS
Stock price prediction using neural networks: A case of Pakistan
MCS-4, Aman-CED, First Floor, Institute of Business Administration, Karachi
This study investigated the performance of two popular neural network architectures, LSTM and GRU, in predicting stock prices. Five-year daily price data of four major Pakistan public companies based on the highest market capital in their respective sectors were used for the analysis. The data was split into training and testing data. The models were also tested using SVM and DT models as performance comparisons with the neural networks. The results of this study showed that the LSTM and GRU models are the most successful in terms of average returns when predicting stock prices because of the ability to capture long-term dependencies in the data, which is beneficial for predicting future stock prices. The SVM and DT models, on the other hand, are much less successful due to their inability to capture long-term dependencies. Based on the findings of this study, it is recommended that investors and traders use the LSTM and GRU models when predicting stock prices. Furthermore, it is recommended that further research be conducted to explore the potential of other models, such as deep learning models, for predicting stock prices.
