Paper: arXiv 2407.14736
Authors: Pascal François, Geneviève Gauthier, Frédéric Godin, Carlos Octavio Pérez Mendoza
Abstract
The recent work of Horikawa and Nakagawa (2024) claims that under a complete market admitting statistical arbitrage, the difference between the hedging position provided by deep hedging and that of the replicating portfolio is a statistical arbitrage. This raises concerns as it entails that deep hedging can include a speculative component aimed simply at exploiting the structure of the risk measure guiding the hedging optimisation problem. We test whether such finding remains true in a GARCH-based market model, which is an illustrative case departing from complete market dynamics. We observe that the difference between deep hedging and delta hedging is a speculative overlay if the risk measure considered does not put sufficient relative weight on adverse outcomes. Nevertheless, a suitable choice of risk measure can prevent the deep hedging agent from engaging in speculation.
Complexity vs Empirical Score
- Math Complexity: 7.5/10
- Empirical Rigor: 8.0/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper employs advanced stochastic calculus, deep reinforcement learning with neural network architectures, and formal definitions of statistical arbitrage (CVaR, VaR). Empirically, it features a backtest-ready setup with a GARCH market model, Monte Carlo simulations, 400k training paths, 100k test paths, and includes a GitHub link for code replication.
Research Flowchart
flowchart TD
A["Research Goal: Is Deep Hedging<br>vs Delta Hedging<br>a Statistical Arbitrage?"] --> B{"Methodology: GARCH Market Model"}
B --> C["Data: Simulated Market Paths"]
C --> D["Process: Optimize Hedging<br>with Risk Measures"]
D --> E{"Risk Measure Choice"}
E -- Insufficient Weight<br>on Adverse Outcomes --> F["Outcome: Deep Hedging acts as<br>Speculative Overlay / Stat Arb"]
E -- Suitable Risk Measure<br>Chosen --> G["Outcome: Speculation Prevented<br>Deep Hedging aligned with Delta"]
F --> H["Key Finding: Risk measure<br>selection dictates whether<br>deep hedging contains speculation"]
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 tests whether the difference between deep hedging and delta hedging constitutes a statistical arbitrage in a GARCH‑based market model and shows how the choice of risk measure affects this.
Source sentence
We test whether such finding remains true in a GARCH-based market model, which is an illustrative case departing from complete market dynamics.
Assumes 5
- Risk‑free rate r and dividend yield q are constant over the horizon. market
Source sentence
the annualized continuously compounded risk‑free rate and dividend yield are assumed to be constant with values set to r = 2.66% and q = 1.77%.
- The hedging objective uses Conditional Value‑at‑Risk (CVaRα) as risk measure. other
Source sentence
we consider the Conditional Value‑at‑Risk (CVaRα) defined as ρ(ξδ T) = E[ξδ T | ξδ T ≥VaRα(ξδ T)], where α ∈(0, 1)
- The cash amount borrowed in the portfolio remains below a fixed bound B>0. execution
Source sentence
cash amount borrowed in the portfolio remains below B > 0 (see François et al. (2024)).
- Training data consist of 400,000 independent simulated paths; test data consist of 100,000 independent paths. data
Source sentence
Agents are trained on training sets of 400,000 independent simulated paths
Show 1 more
- Parameter estimates for the GARCH model are obtained by maximum likelihood on daily S&P 500 data from 2016‑01‑04 to 2020‑12‑31. data
Source sentence
Parameter estimates are obtained through maximum likelihood on a daily time series of the S&P 500 index extending from January 4, 2016, to December 31, 2020.
Data
- Synthetic simulated paths for training synthetic unknown
Source sentence
Agents are trained on training sets of 400,000 independent simulated paths with mini‑batch size of 1,000 ...
- Synthetic simulated paths for testing synthetic unknown
Source sentence
Numerical results are obtained from test sets of 100,000 independent paths.
- S&P 500 daily index series market-prices public
Source sentence
Parameter estimates are obtained through maximum likelihood on a daily time series of the S&P 500 index extending from January 4, 2016, to December 31, 2020.
Costs
Not modeled
Finds 4
- When the CVaR confidence level α is low, the deep‑hedging strategy adds a speculative overlay to delta hedging, creating statistical arbitrage opportunities. statistical
Source sentence
we observe that the difference between deep hedging and delta hedging is a speculative overlay if the risk measure considered does not put sufficient relative weight on adverse outcomes.
- Using a sufficiently high α for CVaR eliminates the statistical‑arbitrage‑like behavior of the differential position between deep and delta hedging. theoretical
Source sentence
... the difference between deep hedging and delta hedging does not exhibit statistical arbitrage‑like behavior
- The objective function of deep hedging must be carefully selected to avoid speculative behavior and achieve sound hedging performance. other
Source sentence
the objective function of the deep hedging problem must be carefully selected to prevent the hedging agent from abandoning its hedging objective and pursuing speculative behavior
- Risk measures that provide no reward for gains (e.g., semi‑RMSE) can prevent speculation but reduce profitability. negative
Source sentence
A possibility could be to use risk measures that do not provide any reward for gains, such as the semi‑RMSE
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 17 related papers). Coverage is partial, so this is neither replication nor refutation.
3 papers in the archive
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.
Cited References (37)
- Relationship between deep hedging and delta hedging: Leveraging a statistical arbitrage strategy (2024)
- Robust risk-aware option hedging — → 2303.15216
- Deep hedging: learning to remove the drift under trading frictions with minimal equivalent near-martingale measures (2021) — link
- Deep hedging of derivatives using reinforcement learning (2020)
- Gamma and vega hedging using deep distributional reinforcement learning (2023)
- Deep hedging (2019)
- Deep hedging of long-term financial derivatives (2021)
- Equal risk pricing of derivatives with deep hedging (2021)
and 29 more
- Deep equal risk pricing of financial derivatives with non-translation invariant risk measures (2023)
- Approximation capabilities of multilayer feedforward networks (1991)
- Empirical deep hedging (2023)
- Deep Hedging with Market Impact — → 2402.13326
- Recent advances in reinforcement learning in finance (2023)
- Deep Reinforcement Learning for Dynamic Stock Option Hedging: A Review (2023)
- Quantum deep hedging — → 2303.16585
- Deep reinforcement learning for option replication and hedging (2020)
- The QLBS Q -Learner goes N u QL ear: F itted Q iteration, inverse RL , and option portfolios (2019)
- Robust deep hedging (2022)
- Reinforcement learning: An introduction (2018)
- Deep hedging: Hedging derivatives under generic market frictions using reinforcement learning (2022)
- Deep learning (2016)
- Deep equal risk pricing of financial derivatives with multiple hedging instruments (2021) — link
- The Pricing of Options and Corporate Liabilities (1973)
- Adam: A method for stochastic optimization (2014) — link
- Deep reinforcement learning for option pricing and hedging under dynamic expectile risk measures (2023)
- Deep hedging under rough volatility (2021)
- Deep reinforcement learning for option pricing and hedging under dynamic expectile risk measures (2023) — link
- No-Transaction Band Network: A Neural Network Architecture for Efficient Deep Hedging (2021) — link
- An empirical analysis of dropout in piecewise linear networks (2013) — link
- Understanding the difficulty of training deep feedforward neural networks (2010)
- Introduction to Stochastic Calculus Applied to Finance (2011)
- On the relation between the expected value and the volatility of the nominal excess return on stocks (1993)
- Statistical arbitrage and securities prices (2003)
- Hedging, P areto optimality, and good deals (2013)
- The GARCH option pricing model (1995)
- Beyond arbitrage: Good-deal asset price bounds in incomplete markets (2000)
- Enhancing Deep Hedging of Options with Implied Volatility Surface Feedback Information — → 2407.21138
4 of 37 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.