Deep reinforcement learning for stock market
| Autor: | Huang, Yu-Hsiang, 黃鈺翔 |
|---|---|
| Rok vydání: | 2019 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 107 Investors put their money into the financial market, hoping to maximize profits by grasping the trend of the market and designing trading strategies at the entry and exit points. However, there are many methods and techniques to judge the market trend, such as judging the market trend through different technical indicators: MA5, MA10, MA20, etc. But even with technical indicators, deciding when to get in and out is a difficult task. Traditional research studies analyze the rise and fall of the market in advance and then further establish trading strategies to invest in the stock market. However, it is very difficult and time consuming. This paper proposes an automatic trading system which combines neural network and reinforcement learning to determine the trading signal and the size of trading position. The system is constructed by LSTM combined with Deep q-learning. This study uses price data of Taiwan stock market, including daily opening price, closing price, highest price, lowest price and trading volume. The profitability of the system is discussed by using the combination of different states of different stocks. The profitability of the system proposed in this study is positive after a long period of testing, which means that the system has a good performance in predicting the rise and fall of stocks. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|