This study delves into the comparison of Informer and LSTM models for forecasting stock market closing prices by incorporating sentiment analysis and technical indicators. The research explores the impact of different feature sets on prediction accuracy and model performance. The study reveals the strengths and limitations of each model configuration through evaluation using metrics like Mean Absolute Error and Mean Absolute Percentage Error. The results highlight the superior performance of the Informer model, particularly when coupled with technical indicators. However, incorporating sentiment analysis showed limited impact on prediction accuracy. These findings shed light on the potential of leveraging deep learning models for stock price prediction and the importance of feature selection in enhancing forecasting capabilities.
Two distinct types of data were gathered for this project. Utilizing the Yahoo Finance API, we retrieved S&P 500 index stock price data spanning from December 1, 2018, to December 31, 2023. The S&P 500 serves as a pivotal stock market index, tracking 500 major publicly traded U.S. companies and standing as a widely monitored benchmark for overall market performance. Daily closing price data was employed for training and prediction purposes.
Clone the repo and run the notebook file or simply open the notebook file in google colab (the name of the notebook represent the feature used in the implementation)
If you have any question or want to use the code, please contact m.wedamerta@innopolis.university.
We appreciate the following github repos for their publicly available code and methdos: