人工智能培训

搜索

深度学习论文:基于模型的动态投资组合优化深度强化学习(Model-based Deep Reinforcement Learning for Dynamic P

[复制链接]
liwei906666 发表于 2019-1-28 11:28:03 | 显示全部楼层 |阅读模式
liwei906666 2019-1-28 11:28:03 187 0 显示全部楼层
深度学习论文:基于模型的动态投资组合优化深度强化学习(Model-based Deep Reinforcement Learning for Dynamic Portfolio  Optimization)动态投资组合优化是根据投资者的回报风险概况,在一些连续交易时段内依次将财富分配给一组资产的过程。使用机器学习自动执行此过程仍然是一个具有挑战性的问题。在这里,我们设计了一个具有自主交易代理的深度强化学习(RL)架构,以便根据全球目标,自主权定期制定投资决策和行动。特别是,在不依赖纯粹的无模型RL代理的情况下,使用由注入预测模块(IPM),生成性对抗数据增强模块(DAM)和行为克隆模块(BCM)组成的新型RL架构来润湿我们的交易代理。我们基于模型的方法适用于策略上或非策略RL算法。我们进一步设计了后端测试和执行引擎,它实时与RL代理进行交互。使用历史{\ em real}金融市场数据,我们模拟具有实际约束的交易,并证明我们提出的模型与基线交易策略和先前工作中的无模型RLagent相比具有鲁棒性,盈利性和风险敏感性。
Dynamic portfolio optimization is the process of sequentially allocatingwealth to a collection of assets in some consecutive trading periods, based oninvestors' return-risk profile.Automating this process with machine learningremains a challenging problem.Here, we design a deep reinforcement learning(RL) architecture with an autonomous trading agent such that, investmentdecisions and actions are made periodically, based on a global objective, withautonomy.In particular, without relying on a purely model-free RL agent, wetrain our trading agent using a novel RL architecture consisting of an infusedprediction module (IPM), a generative adversarial data augmentation module(DAM) and a behavior cloning module (BCM).Our model-based approach works withboth on-policy or off-policy RL algorithms.We further design the back-testingand execution engine which interact with the RL agent in real time.Usinghistorical {\em real} financial market data, we simulate trading with practicalconstraints, and demonstrate that our proposed model is robust, profitable andrisk-sensitive, as compared to baseline trading strategies and model-free RLagents from prior work.深度学习论文:基于模型的动态投资组合优化深度强化学习(Model-based Deep Reinforcement Learning for Dynamic Portfolio  Optimization) IX66P5tyUOyX5yl8.jpg
URL地址:https://arxiv.org/abs/1901.08740     ----pdf下载地址:https://arxiv.org/pdf/1901.08740    ----深度学习论文:基于模型的动态投资组合优化深度强化学习(Model-based Deep Reinforcement Learning for Dynamic Portfolio  Optimization)
回复

使用道具 举报

您需要登录后才可以回帖 登录 | 立即注册

本版积分规则 返回列表 发新帖

liwei906666当前离线
新手上路

查看:187 | 回复:0

快速回复 返回顶部 返回列表