Financial Trading as a Game: A Deep Reinforcement Learning Approach

arxiv:1807.02787

Original source · Paper JSON

Finance · 2018-07-08

Cites (1) · Cited by (0)

Complete cleaned Markdown · Source Markdown including bibliography

Read in parts

1. Part 1
2. Part 2

Contents

• Financial Trading as a Game: A Deep Reinforcement Learning Approach
• Abstract
• keywords
• 1 Introduction
• 1.1 The Financial Trading Task
• 1.2 Motivation
• Definition 1 .
• 1.3 Challenges
• 1.4 Contributions
• 2 Method
• 2.1 Data Preparation and Feature Extraction
• 2.2 Financial Trading MDP
• 2.2.1 State Space ∈ ℝ 198 \in\mathbb{R}^{198}
• 2.2.2 Action Space
• 2.2.3 Reward Function
• 2.3 Fully Exploit with Action Augmentation
• 2.4 Model Architecture
• 2.4.1 Weight Initialization
• 2.5 Training Scheme
• 2.5.1 Modified Training Scheme
• 2.5.2 A Complete Online Learning Algorithm
• 3 Experiment
• 3.1 Hyperparameters
• 3.2 Simulation Result
• 3.3 Effect of the Spread
• 3.4 Effectiveness of Action Augmentation
• 4 Conclusion
• 4.1 Achievements
• 4.2 Future Work
• Appendix