Full Text Available
Note: Clicking the button above will open the full text document at the original institutional repository in a new window.
Using Hybrid Machine learning Models for Stock Price Forecasting and Trading.
Trading stocks of publicly traded companies in stock markets is a challenging topic since investors are researching what tools can be used to maximize their profits while minimizing risks, which encouraged all researchers to research and test different methods to reach such a goal. As a result, the...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Published: |
AUC Knowledge Fountain
2024
|
| Subjects: | |
| Tags: |
No Tags, Be the first to tag this record!
|
| Summary: | Trading stocks of publicly traded companies in stock markets is a challenging topic since investors are researching what tools can be used to maximize their profits while minimizing risks, which encouraged all researchers to research and test different methods to reach such a goal. As a result, the use of both fundamental analysis and technical analysis started to evolve to support traders in buying and selling stocks. Recently, the focus increased on using Machine learning models to predict stock prices and algorithmic trading as currently there is a huge amount of data that can be processed and used to forecast and trade stocks.
The focus of this paper is to use four machine learning models to forecast next day stock prices and trade stocks accordingly. The models used are LSTM model, W-LSTM model in which Wavelet analysis is used to remove the noise of the time series data and provide new coefficients to the LSTM model, LSTM-ARO where ARO as an optimization algorism will select the best hyperparameters for the LSTM model and W-LSTM-ARO model. The model’s prediction accuracy is evaluated by MSE, MAE & R2, then the models are tested in terms of profit generation.
The companies studied in this paper are six companies listed on the New York Stock Exchange (NYSE): Apple (APPL), Microsoft (MSFT), Exxon Mobile (XOM), General Electric (GE), AT&T (T) and Procter and Gamble (PG).
The study concludes that there is no correlation between models with high prediction accuracy and the ability of the model to generate profits. There is no best model that will fit in trading all stocks. Finally, the profits generated by W-LSTM & W-LSTM-ARO are higher the ones generated by LSTM & LSTM-ARO models. When the profits generated by W-LSTM & W-LSTM-ARO are compared with the buy & hold, RSI & EMA trading strategies it was concluded that W-LSTM & W-LSTM-ARO models are able to generate profits when the stock is in a downtrend. |
|---|