Loading...

Prediction of Stock Market Based on Corporate Financial Reports Using Deep Learning

Shafiei Masoleh, Hamed | 2021

322 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 53831 (19)
  4. University: Sharif University of Technology
  5. Department: Computer Engineering
  6. Advisor(s): Sameti, Hossein
  7. Abstract:
  8. Creating tools for automating trade or creating advisory tools have great importance for stock markets. Regarding stock markets, information varies in type e.g. financial disclosures, news, price history, audit reports, etc. Aforementioned information and data's variance, volume, and the high number of factors affecting the stock price, make the stock market hard to predict all together. Therefore, predictions are usually limited to a subset of data. The goal of this research is to take advantage of the newest language processing techniques in order to analyze financial disclosure documents and predict their effect on their related stock price. Financial disclosures usually have a longer text. A dataset of German companies' financial disclosures in English was used as the primary dataset. In this research, three approaches were suggested: 1- Truncating the documents and using a pre-trained RoBERTa model. 2- Using Longformer model. 3- Summarizing documents and using RoBERTa model. For the first approach, comparing with the baseline model, accuracy was lifted slightly from 0.580 to 0.581 but balanced accuracy was decreased from 0.571 to 0.564. As for the second approach, accuracy and balanced accuracy were 0.597 and 0.592 which increased the baseline's outcome by 0.012 and 0.021 respectively. The third approach was not implementable due to a lack of hardware computational resources. In addition to accuracy and balanced accuracy, precision, recall, and F1 score were reported which for the first approach the results were 0.609, 0.650, and 0.629 and for the second approach, the results were 0.630, 0.642, and 0.636 respectively. Therefore, improvement can be achieved by utilizing pre-trained transformer-based models. Moreover, there is still space for improvement
  9. Keywords:
  10. Natural Language Processing ; Neural Networks ; Deep Learning ; Price Forecasting ; Stock Price ; Stock Prediction

 Digital Object List

 Bookmark

...see more