|
JAIST Repository >
d. 融合科学系 >
d11. 会議発表論文 >
d11-1. 会議発表論文 >
このアイテムの引用には次の識別子を使用してください:
http://hdl.handle.net/10119/20016
|
| タイトル: | Optimal execution strategy using Deep Q-Network with heuristics policy |
| 著者: | Ogawa, Tatsuyoshi
Nakagawa, Kei
Ikeda, Kokolo |
| キーワード: | optimal execution problem
DQN
DDQN
TWAP |
| 発行日: | 2024-07-06 |
| 出版者: | Institute of Electrical and Electronics Engineers (IEEE) |
| 誌名: | 2024 16th IIAI International Congress on Advanced Applied Informatics (IIAI-AAI) |
| 開始ページ: | 456 |
| 終了ページ: | 461 |
| DOI: | 10.1109/IIAI-AAI63651.2024.00089 |
| 抄録: | The optimal execution problem involves planning a stock execution strategy that minimizes trading costs for a specific quantity of stock over a certain timeframe. To tackle this problem, advanced techniques like Deep Reinforcement Learning (DRL), especially the Deep Q-Network (DQN) which employs deep learning to approximate the Q value function, have been introduced to identify the most efficient execution strategies. However, DRL methods face challenges such as learning instability and the extensive data requirements. Therefore, we propose to use prioritized experience replay and to incorporate a strategy derived from the insights of the financial field into the DQN during learning process. Particularly, we introduce a time-weighted average price (TWAP) strategy that has been proven to be optimal under specific conditions as a heuristic policy. This approach is expected to be able to enhance the stability and performance of policy learning. We have conducted numerical experiments in various noise-prone environments to assess the effectiveness of our approach. The findings indicate that our proposed method consistently outperforms conventional benchmarks by reducing costs in all tested environments. |
| Rights: | This is the author's version of the work. Copyright (C) 2024 IEEE. 2024 16th IIAI International Congress on Advanced Applied Informatics (IIAI-AAI), Takamatsu, Japan, pp. 456-461. DOI: https://doi.org/10.1109/IIAI-AAI63651.2024.00089. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works. |
| URI: | http://hdl.handle.net/10119/20016 |
| 資料タイプ: | author |
| 出現コレクション: | d11-1. 会議発表論文 (Conference Papers)
|
このアイテムのファイル:
| ファイル |
記述 |
サイズ | 形式 |
|---|
| T-IKEDA-K-0930-5.pdf | | 1097Kb | Adobe PDF | 見る/開く |
|
当システムに保管されているアイテムはすべて著作権により保護されています。
|
