Loading...

Providing a Trading Framework for Financial Markets using Machine Learning Techniques

Azizi, Mohammad Javad | 2024

0 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 57613 (01)
  4. University: Sharif University of Technology
  5. Department: Industrial Engineering
  6. Advisor(s): Rafiee, Majid
  7. Abstract:
  8. In recent years, investment in financial markets has seen exponential growth in the number of investors and the capital they have invested. Investors enter these markets with the goal of maximizing their transaction profits. Therefore, they are looking for a way to predict the future of the market. On the other hand, numerous factors influence stock prices and market trends, making the prediction of these markets a highly complex task. In recent years, with the remarkable advancement of artificial intelligence, analysts’ ability to predict stock trends has greatly improved. In this research, we attempt to predict stock trends using financial market concepts and machine learning models. To this end, by examining 26 technical indicators and creating innovative trading strategies, we generate over a thousand features in our dataset. We form our target variable to predict the upward or downward trend of stocks based on local price floors and ceilings. Using 9 traditional machine learning models and neural networks such as logistic regression, support vector machines, and artificial neural networks, we predict the target variable to issue buy or sell signals for each trading day. Based on performance evaluation metrics such as F1-score and trading profits, we compare 135 different scenarios and identify the strengths and weaknesses of each. The results show that the presented models are capable of yielding up to 4.7 times the return of the buy-and-hold strategy. In the best case, we achieved a 78% F1-score and a 127% return in less than one trading year. Subsequently, using a convolutional neural network model, we processed the image of the stock chart and achieved satisfactory results. Ultimately, it was determined that the innovations presented in creating the dataset and using machine learning models were beneficial and effective
  9. Keywords:
  10. Financial Market ; Market Prediction ; Stock Prediction ; Machine Learning ; Neural Network ; Deep Learning

 Digital Object List

 Bookmark

No TOC