Paper: arXiv 2402.13326

Authors: Andrei Neagu, Frédéric Godin, Clarence Simard, Leila Kosseim

Abstract

Dynamic hedging is the practice of periodically transacting financial instruments to offset the risk caused by an investment or a liability. Dynamic hedging optimization can be framed as a sequential decision problem; thus, Reinforcement Learning (RL) models were recently proposed to tackle this task. However, existing RL works for hedging do not consider market impact caused by the finite liquidity of traded instruments. Integrating such feature can be crucial to achieve optimal performance when hedging options on stocks with limited liquidity. In this paper, we propose a novel general market impact dynamic hedging model based on Deep Reinforcement Learning (DRL) that considers several realistic features such as convex market impacts, and impact persistence through time. The optimal policy obtained from the DRL model is analysed using several option hedging simulations and compared to commonly used procedures such as delta hedging. Results show our DRL model behaves better in contexts of low liquidity by, among others: 1) learning the extent to which portfolio rebalancing actions should be dampened or delayed to avoid high costs, 2) factoring in the impact of features not considered by conventional approaches, such as previous hedging errors through the portfolio value, and the underlying asset’s drift (i.e. the magnitude of its expected return).

Complexity vs Empirical Score

  • Math Complexity: 8.0/10
  • Empirical Rigor: 4.0/10
  • Quadrant: Lab Rats — high math complexity, limited empirical testing

Why this score: The paper heavily employs advanced mathematics, including stochastic control, convex optimization, and deep reinforcement learning theory, with detailed derivations and formalisms. While it uses simulations to compare against delta hedging, there is no mention of real-world backtests, actual limit order book data, or statistical metrics like Sharpe ratios on live data, placing it in the ‘Lab Rats’ quadrant.

Research Flowchart

  flowchart TD
  A["Research Goal<br>Optimize hedging with<br>market impact using DRL"] --> B{"Methodology"}
  
  B --> C["Data & Inputs"]
  C --> D["Option & Stock Prices<br>Convex Market Impact<br>Liquidity Constraints"]
  
  B --> E["Computational Process"]
  E --> F["Deep Reinforcement Learning<br>Environment: Sequential Trading<br>Action: Hedge Ratio"]
  
  F --> G["Optimization<br>Minimize Costs vs Tracking Error<br>via Reward Function"]
  
  G --> H["Key Findings & Outcomes"]
  H --> I["DRL Model Outperforms<br>Delta Hedging in Low Liquidity"]
  H --> J["Optimal Policy Learned:<br>Dampen/Delay Actions to<br>Avoid High Costs"]
  H --> K["Incorporates Feature Impact:<br>Previous Errors & Asset Drift"]
  
  D --> F
  G --> H

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 a novel deep reinforcement learning model for dynamic option hedging that incorporates convex market impacts and impact persistence.

Source sentence
In this paper, we propose a novel general market impact dynamic hedging model based on Deep Reinforcement Learning (DRL) that considers several realistic features such as convex market impacts, and impact persistence through time.

Assumes 6

  • Price dynamics follow a geometric Brownian motion with constant drift µ and volatility σ. model
    Source sentence
    we used the geometric Brownian motion (GBM) to simulate the underlying asset prices St at each time step t: St = St−1 exp (µ −σ2/2) δt + σ
  • Market impact is convex and can be parameterized by illiquidity parameters α and β. market
    Source sentence
    The following three sets of market impact parameters are considered: α = β = 1 (infinite liquidity); α = 1.01, β = 0.99 (high liquidity); α = 1.02, β = 0.98 (low liquidity).
  • Impact persistence is modeled by parameters λa and λb; the no‑persistence case sets λa = λb = ∞. market
    Source sentence
    The first set of analyses assumes no impact persistence (i.e. λa = λb = ∞), which means that market transactions have only an immediate impact.
  • Risk‑free rate is set to zero for simplicity. model
    Source sentence
    For simplicity, the risk‑free rate is set to r = 0.
Show 2 more
  • The hedging portfolio value Vt and the underlying drift µ are included as state variables. execution
    Source sentence
    The hedging portfolio value Vt and the underlying asset drift (i.e. expected return) parameter µ materially impact optimal positions.
  • Training uses a policy‑gradient algorithm with Adam‑optimized learning rates. statistical
    Source sentence
    the learning rates sequence {ηj}j≥1 contains small positive real numbers determined by the Adam optimizer [26]

Data

  • Synthetic GBM price series synthetic unknown
    Source sentence
    we used the geometric Brownian motion (GBM) to simulate the underlying asset prices St at each time step t

Costs

Modeled Costs are part of the reported headline results.

A proportional transaction cost rate k is calibrated so that buying one share at initiation with fees k matches the DRL model revenue.

Source sentence
In all experiments, the parameter k is set so that the revenue of buying one share of the underlying asset at the initiation date with the proportional fees k is equal to that of the DRL model F b(0, 0, 1).

Finds 5

  • The DRL model outperforms delta‑hedging baselines in low‑liquidity settings by learning to dampen or delay rebalancing actions. performance
    Source sentence
    Results show our DRL model behaves better in contexts of low liquidity by, among others: 1) learning the extent to which portfolio rebalancing actions should be dampened or delayed to avoid high costs
  • Optimal positions depend on the hedging portfolio value Vt, unlike delta‑hedging which ignores Vt. theoretical
    Source sentence
    our optimal DRL strategy depends on the hedging portfolio value at time t (Vt) whereas both delta hedging strategies ignore Vt in their decisions
  • The DRL policy implicitly incorporates the underlying drift µ, whereas traditional delta hedging does not. theoretical
    Source sentence
    our DRL model’s optimal hedging positions do implicitly take into account the drift
  • Large illiquidity parameters α,β encourage delaying corrections, but large impact‑persistence parameters λa,λb counteract this effect. theoretical
    Source sentence
    while large values of illiquidity parameters α, β incentivize partially delaying the correction of inadequate hedging positions
  • When impact persistence is absent, market impacts vanish immediately after a trade (At = Bt = 0). statistical
    Source sentence
    The absence of impact persistence is reflected by setting λa = λb = ∞ and thus At = Bt = 0 for all time

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 24 related papers). Coverage is partial, so this is neither replication nor refutation.

Cited by 7 papers in the archive (2 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.

Cited References (49)

  1. LaTeX Compatibility (2021)
  2. Handbook of mathematical functions with formulas, graphs, and mathematical tables (1948)
  3. Tensorflow: Large-scale machine learning on heterogeneous distributed systems (2016) — link
  4. Nonquadratic local risk-minimization for hedging contingent claims in incomplete markets (2011)
  5. Optimal execution of portfolio transactions (2001)
  6. A profitable modification to global quadratic hedging (2019)
  7. Mean–variance hedging under transaction costs (2007)
  8. The pricing of options and corporate liabilities (1973)
and 41 more
  1. Deep hedging (2019)
  2. On E sscher transforms in discrete finance models (1998)
  3. Deep equal risk pricing of financial derivatives with non-translation invariant risk measures (2023)
  4. Equal risk pricing of derivatives with deep hedging (2021)
  5. Deep hedging of long-term financial derivatives (2021)
  6. Optimal execution with limit and market orders (2015)
  7. Local risk-minimization with multiple assets under illiquidity with applications in energy markets (2018)
  8. A Stochastic Model for Order Book Dynamics (2010)
  9. Equal risk pricing and hedging of financial derivatives with convex risk measures (2020) — link
  10. Mean-variance hedging in continuous time (1991)
  11. A discrete time equivalent martingale measure (1998)
  12. Option pricing by Esscher transforms (with discussion) (1994)
  13. Hedging by sequential regression: An introduction to the mathematics of option trading (1989)
  14. Optimal hedging when the underlying asset follows a regime-switching M arkov process (2014)
  15. The minimal entropy martingale measure and the valuation problem in incomplete markets (2000)
  16. Understanding the difficulty of training deep feedforward neural networks (2010)
  17. Minimizing CV a R in global dynamic hedging with transaction costs (2016)
  18. A Closed-Form Solution for the Global Quadratic Hedging of Options under Geometric Gaussian Random Walks (2019)
  19. Option pricing and hedging with execution costs and market impact (2017)
  20. Equal risk pricing under convex trading constraints (2017)
  21. Martingales and arbitrage in multiperiod securities markets (1979)
  22. Options, futures, and other derivative securities (1993)
  23. Gamma and vega hedging using deep distributional reinforcement learning (2023)
  24. Robust Risk-Aware Option Hedging — → 2303.15216
  25. Optimal replication of contingent claims under transaction costs (1989)
  26. Multilayer feedforward networks are universal approximators (1989)
  27. Adam: A method for stochastic optimization (2014) — link
  28. Local risk-minimization under transaction costs (1998)
  29. Option pricing and replication with transaction costs (1985)
  30. An example of indifference prices under exponential preferences (2004)
  31. Double deep Q -learning for optimal execution (2021)
  32. Optimal hedging in discrete time (2013)
  33. Statistical methods for financial engineering (2016)
  34. Resilient price impact of trading and the cost of illiquidity
  35. L 'e vy processes in finance: pricing financial derivatives (2003)
  36. Variance-optimal hedging in discrete time (1995)
  37. A guided tour through quadratic hedging approaches (2001)
  38. General model for limit order books and market orders (2014)
  39. Deep learning for limit order books (2019)
  40. Universal features of price formation in financial markets: perspectives from deep learning (2019)
  41. A mean-field control problem of optimal portfolio liquidation with semimartingale strategies (2023)

1 of 49 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.