Robust financial trading system with Deep Q Network (DQN)
Issued Date
2017
Issued Date (B.E.)
2560
Available Date
Copyright Date
Resource Type
Series
Edition
Language
eng
File Type
application/pdf
No. of Pages/File Size
125 leaves
ISBN
ISSN
eISSN
Other identifier(s)
b203303
Identifier(s)
Access Rights
Access Status
Rights
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
Rights Holder(s)
Physical Location
National Institute of Development Administration. Library and Information Center
Bibliographic Citation
Citation
Sutta Sornmayura (2017). Robust financial trading system with Deep Q Network (DQN). Retrieved from: http://repository.nida.ac.th/handle/662723737/4077.
Title
Robust financial trading system with Deep Q Network (DQN)
Alternative Title(s)
Author(s)
Advisor(s)
Editor(s)
item.page.dc.contrubutor.advisor
Advisor's email
Contributor(s)
Contributor(s)
Abstract
Forex trading is one of the most attractive areas in finance. However, developing the profitable trading system is not an easy task because it requires extensive knowledge in several areas such as quantitative analysis, financial skills, and computer programming. Trading system expert, as a human, also bring in their own bias to develop the system. The trading system developers will prefer some markets over others, prefer some indicators over others, and prefer some trading time frame over others. Moreover, developing the trading system will also be prone to data-snooping and look-ahead bias. Developing trading system is the never-ending task and requires numerous experiments with several parameters.
Random walk and EMH theories support the assumption for choosing buy-and-hold as the best alternative when choosing the strategy for trading. However, there are numerous studies which contradict both theories. Those studies support the idea of using technical analysis as a predictive tool to find the hidden profitable pattern in the market. However, technical analysis is also prone to bias of the users as well.
The problem of developing the robust financial trading system is challenging. In terms of developing cost, time and effort. It must be a new method to efficiently develop the trading system. Simultaneously, this method should eliminate all biases from system developers.
The most attractive way to develop the system is to use cutting-edge technology such as artificial intelligence (AI) technology. This new method of developing the trading strategy needs to benchmark with buy-and-hold (Random walk and EMH assumption) and with the trading experts who are commodity trading advisor (CTA).
This study tried to compare the performance of AI to buy-and-hold strategy and performance of AI to the expert trader. The tested markets were Forex (EURUSD, USDJPY) and Gold (XAUUSD) market, data obtaining from Dukascopy Bank SA Switzerland (15 years data). Both hypotheses were tested with Paired t-Test at the significance level of 0.05. The findings showed that AI could significantly beat buy-and-hold strategy for FOREX in both 2 currency pairs (EURUSD, USDJPY), and AI could also significantly outperform Commodity Trading Advisor (CTA) for trading EURUSD. However, AI could not significantly outperform CTA for USDJPY trading. For Gold (XAUUSD) market, AI could not significantly outperform buy-and-hold and CTA. Limitation, contribution, and further research were also recommended.
Table of contents
Description
Thesis (Ph.D.(Management) )--National Institute of Development Administration, 2017