Loading...

Stock Price Prediction Using Machine Learning Models

Mohammadi, Khashayar | 2024

0 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 58409 (01)
  4. University: Sharif University of Technology
  5. Department: Industrial Engineering
  6. Advisor(s): Khedmati, Majid
  7. Abstract:
  8. Stock price prediction poses a significant challenge for traders, investors, and researchers due to the inherently irregular and chaotic nature of financial markets. To gain a competitive edge—defined as any factor that enhances the rate of capital growth over competitors—investors employ various techniques, with price forecasting being paramount. With the advancement of data mining, machine learning algorithms have emerged as powerful tools for predicting future stock prices. While previous studies have indicated the superior performance of neural network models over classical machine learning approaches, this research investigates the potential for further accuracy improvements through model hybridization. The dataset for this study comprises historical time-series stock price data over a specified period, augmented with three widely used technical indicators: Moving Average Convergence Divergence (MACD), Relative Strength Index (RSI), and the Stochastic Oscillator. To test the hypothesis that hybrid models offer improved predictive accuracy, this thesis proposes a novel hybrid model combining a Neural Network (NN) and a Random Forest (RF). For baseline comparison, several models were also implemented and assessed individually: a time-series ARIMA model, Support Vector Machine (SVM), Random Forest (RF), and a Long Short-Term Memory (LSTM) neural network. The performance of each model was evaluated using the Mean Squared Error (MSE) metric. The results demonstrate that the Support Vector Machine yielded the lowest accuracy. Conversely, the proposed hybrid NN-RF model achieved the highest accuracy, significantly outperforming all baseline models. Crucially, the hybrid model demonstrated a marked improvement in predictive accuracy when compared to the performance of both the standalone Neural Network and the standalone Random Forest models, confirming the synergistic benefits of model combination in this domain
  9. Keywords:
  10. Algorithmic Trading ; Neural Network ; Random Forest Algorithm ; Stock Price Prediction ; Hybrid Machine Learning Model ; Machine Learning in Finance

 Digital Object List

 Bookmark

No TOC