Paper: arXiv 2402.15994

Authors: Qishuo Cheng, Le Yang, Jiajian Zheng, Miao Tian, Duan Xin

Abstract

Portfolio management issues have been extensively studied in the field of artificial intelligence in recent years, but existing deep learning-based quantitative trading methods have some areas where they could be improved. First of all, the prediction mode of stocks is singular; often, only one trading expert is trained by a model, and the trading decision is solely based on the prediction results of the model. Secondly, the data source used by the model is relatively simple, and only considers the data of the stock itself, ignoring the impact of the whole market risk on the stock. In this paper, the DQN algorithm is introduced into asset management portfolios in a novel and straightforward way, and the performance greatly exceeds the benchmark, which fully proves the effectiveness of the DRL algorithm in portfolio management. This also inspires us to consider the complexity of financial problems, and the use of algorithms should be fully combined with the problems to adapt. Finally, in this paper, the strategy is implemented by selecting the assets and actions with the largest Q value. Since different assets are trained separately as environments, there may be a phenomenon of Q value drift among different assets (different assets have different Q value distribution areas), which may easily lead to incorrect asset selection. Consider adding constraints so that the Q values of different assets share a Q value distribution to improve results.

Complexity vs Empirical Score

  • Math Complexity: 4.0/10
  • Empirical Rigor: 7.0/10
  • Quadrant: Street Traders — empirical and applied, lower math complexity

Why this score: The paper uses standard deep reinforcement learning (DQN) without advanced derivations, keeping math complexity moderate, but it includes detailed empirical backtesting with specific datasets, performance tables, and transaction cost analysis, demonstrating high empirical readiness.

Research Flowchart

  flowchart TD
  A["Research Goal:<br>Optimize Portfolio Management &<br>Risk Assessment in Digital Assets"] --> B["Key Methodology: Deep Q-Network DQN"]
  B --> C{"Computational Process:<br>Single Asset Training"}
  C --> D["Outcome:<br>Performance Exceeds Benchmark"]
  C --> E["Issue:<br>Q-value Drift Across Assets"]
  E --> F["Proposed Improvement:<br>Constrain Q-values to<br>Shared Distribution"]
  F --> G["Final Outcome:<br>Enhanced DRL for Portfolio Management<br>Adapting to Financial Complexity"]
  D --> G

Evidence Card

What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card

Proposes

The paper proposes introducing the Deep Q‑Network (DQN) algorithm for asset portfolio management and trading decisions.

Source sentence
In this paper, the DQN algorithm is introduced into asset management portfolios in a novel and straightforward way, and the performance greatly exceeds the benchmark, which fully proves the effectiveness of the DRL algorithm in portfolio management.

Assumes 1

  • The learning process assumes that data samples are independent and identically distributed. statistical
    Source sentence
    This correlation disrupts the independence assumption, leading to significant fluctuations in the learned value function model.

Data

  • Historical data of 500 US stocks market-prices unknown
    2010-01-01 to 2021-06-30 · 500 US stocks
    Source sentence
    The algorithm selected the historical data of 500 US stocks from 2010-01-01 to 2021-06-30 for testing.

Costs

Modeled Costs are part of the reported headline results.

Finds 3

  • The DQN strategy outperforms benchmark strategies in 36 out of 48 experimental groups. performance
    Source sentence
    Among 48 groups of experiments, DQN was the best in 36 groups of experiments.
  • DQN achieves higher out‑of‑sample cumulative returns than buy‑and‑hold, momentum, and reversion baselines across portfolio sizes. performance
    Source sentence
    Out-of-sample cumulative returns at the end of the test period for each level of
  • The DQN strategy performs well in normal and recovery market phases but poorly during market decline phases. performance
    Source sentence
    The DQN strategy performed optimally in both the normal phase and the recovery phase, but in the market decline phase, DQN performed poorly, possibly because there was no similar situation in the training data.

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 25 related papers). Coverage is partial, so this is neither replication nor refutation.

Cited by 1 paper in the archive (1 by independent authors)

Code

No repository by the authors found in the paper.

Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.