Loading...

A Machine Learning and Time-Frequency Domain Combined Approach for Improving Stock Portfolio Management

Dezhkam, Arsalan | 2022

138 Viewed
  1. Type of Document: Ph.D. Dissertation
  2. Language: Farsi
  3. Document No: 55577 (19)
  4. University: Sharif University of Technology
  5. Department: Computer Engineering
  6. Advisor(s): Manzuri, Mohammad Taghi
  7. Abstract:
  8. Price prediction in financial markets is an exciting problem for a vast majority of groups and people; however, investment portfolio managers and owners are always looking for holistic predic-tion approaches and tools having high functional accurate metrics. Strictly speaking, players in fi-nancial markets are always in search of methods and toolboxes since they need to overcome the un-certainty of their buy, sell, or hold decisions in order to reduce the investment risk. In this research, we have tried to deal with the stock price prediction problem as an asset pricing problem and find a novel approach to push forward the state-of-the-art of the problem based on the fundamental pric-ing theory. The study is conducted to find the most pertinent but non-collinear features along with the macroeconomic factors having the highest amount of information for pricing the stock assets. The proposed model is a combination of Hilbert-Huang Transform, HHT, as the feature engineering part, and the extreme gradient boost, XGBoost, as the Close price trend classifier. The classification output is a sequence of ups and downs used for optimizing the portfolio weights of the stocks with the best trading performance. The performance of the portfolios optimized under this study proves that our novel combination of HHT with classification performs 211% better than forming the port-folio using raw financial data
  9. Keywords:
  10. Financial Time Series ; Deep Learning ; Optimal Portfolio ; Machine Learning ; Time Frequency Transform ; Market Trend Prediction ; Portfolio Management

 Digital Object List

 Bookmark

No TOC