| Summary: | In the financial market, predicting stock price movement has always been a challenge.
Many investors and analysts use different analyzing techniques, trying to predict the volatile
market. This project presents a technological approach to short-term stock price prediction
using Long Short-Term Memory (LSTM) neural networks. The primary objective of this
research is to develop and evaluate the accuracy of an LSTM-based machine learning model
in predicting short-term stock movement. The model was trained and tested using a several
datasets, comprising of stock indexes and individual stocks from the US, China and Hong
Kong.
The LSTM model was configured and implemented using the Keras API, with
GridSearchCV hyperparameter tuning to optimize performance. The model’s accuracy was
evaluated using Mean Squared Error (MSE) and R-Squared (R2) metrics. Additionally,
technical analysis was conducted to provide further information into the stock movement.
The results demonstrate great performance in predicting movement of stocks. However, it is
emphasized that the model’s prediction is a gauge and should be used in conjunction with
other analytical techniques, such as technical analysis, to make well-informed decisions.
The findings of this study highlight the potential of machine learning like LSTM to
predict future stock price movements and offer valuable indicators for investors. Future work
may include refining the model, exploring additional data sources and different analytical
techniques such as fundamental analysis to improve prediction performances.
|
|---|