Paper: arXiv 2507.17162
Authors: Patrick Chan, Ronnie Sircar, Iosif Zimbidis
Abstract
We consider a dynamic portfolio optimization problem that incorporates predictable returns, instantaneous transaction costs, price impact, and stochastic volatility, extending the classical results of Garleanu and Pedersen (2013), which assume constant volatility. Constructing the optimal portfolio strategy in this general setting is challenging due to the nonlinear nature of the resulting Hamilton-Jacobi-Bellman (HJB) equations. To address this, we propose a multi-scale volatility expansion that captures stochastic volatility dynamics across different time scales. Specifically, the analysis involves a singular perturbation for the fast mean-reverting volatility factor and a regular perturbation for the slow-moving factor. We also introduce an approximation for small price impact and demonstrate its numerical accuracy. We formally derive asymptotic approximations up to second order and use Monte Carlo simulations to show how incorporating these corrections improves the Profit and Loss (PnL) of the resulting portfolio strategy.
Complexity vs Empirical Score
- Math Complexity: 9.0/10
- Empirical Rigor: 3.0/10
- Quadrant: Lab Rats — high math complexity, limited empirical testing
Why this score: The paper involves highly advanced mathematical techniques including Hamilton-Jacobi-Bellman equations, singular/regular perturbation theory for multiscale stochastic volatility, and second-order asymptotic expansions. While it uses Monte Carlo simulations to demonstrate theoretical improvements, it lacks code, backtests, statistical metrics, or implementation details required for empirical readiness.
Research Flowchart
flowchart TD A["Research Goal: Optimal Portfolio with Stochastic Volatility & Price Impact"] --> B["Methodology: Multi-Scale Volatility Expansion<br>Fast (Singular) & Slow (Regular) Perturbations"] B --> C["Data: Simulated Price Dynamics<br>Inputting Predictable Returns, Volatility Regimes, Impact Costs"] C --> D["Computational Process: Solving Nonlinear HJB Equation<br>via Asymptotic Approximations"] D --> E["Outcome: Asymptotic Portfolio Strategy<br>Incorporating Impact & Volatility Corrections"] E --> F["Validation: Monte Carlo Simulations"] F --> G["Key Findings: <br>1. Improved PnL from Multi-Scale Corrections<br>2. Formal 2nd-Order Asymptotics"]
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 multi‑scale volatility expansion using singular and regular perturbations for fast and slow stochastic volatility factors, and introduces a small‑price‑impact approximation to obtain tractable optimal trading strategies.
Source sentence
To address this, we propose a multi-scale volatility expansion that captures stochastic volatility dynamics across different time scales. Specifically, the analysis involves a singular perturbation for the fast mean-reverting volatility factor and a regular perturbation for the slow-moving factor.
Assumes 6
- The fast and slow volatility factors evolve on separate time scales characterized by small parameters ε (fast) and δ (slow). model
Source sentence
The parameters ε and δ, when small, characterize the fast and slow variation of the volatility factors Y and Z, respectively.
- Price impact is linear in the change of portfolio position and decays exponentially over time. model
Source sentence
In particular, it is linear in the change in portfolio position and decays exponentially over time. Thus, it can be modeled as: dlt = −β lt dt + λ dqt = (λut −βlt) dt, (3) where β and λ are positive constants.
- The decay rate β and liquidity parameter λ of the price‑impact dynamics are constant throughout the analysis. model
Source sentence
However, for simplicity, we assume these parameters remain constant throughout our analysis.
- The scaling parameter θ used for the small‑price‑impact approximation is assumed to be small. model
Source sentence
Thus, we introduce the scalling paramter θ into the dynamics of the price impact process as: dlt = −θβlt dt + θλ dqt, where θ is assumed to be small.
Show 2 more
- The risk‑free interest rate is zero. market
Source sentence
The trader follows a self-financing trading strategy to maximize the financial gain from their position in the stock: Πs = Z s 0 qt dSt.
- Brownian motions driving the factors are independent of the price process. model
Source sentence
We also assume that the Brownian motions W (0) t , W (1) t , W (2) t are independent of Bt.
Data
Synthetic data only.
Costs
Modeled
Instantaneous transaction costs and linear price impact are modeled with parameters β and λ (β decay rate, λ market liquidity).
Source sentence
In particular, it is linear in the change in portfolio position and decays exponentially over time. Thus, it can be modeled as: dlt = −β lt dt + λ dqt = (λut −βlt) dt, (3) where β and λ are positive constants.
Finds 3
- In Monte Carlo simulations the corrected strategy that incorporates fast mean‑reverting stochastic volatility improves PnL by an average of 53.27 basis points relative to the constant‑volatility strategy. performance
Source sentence
In particular, the corrected strategy leads to a substantial improvement in PnL, with an average increase of 53.27 basis points relative to the constant volatility strategy, when Y0 = 0.4.
- The second‑order correction has a significant impact on portfolio performance, leading to substantial PnL improvement. performance
Source sentence
As observed in Figure 4, the second-order correction has a significant impact on portfolio perfor-mance.
- Under fast‑scale dynamics, traders should deleverage when current volatility exceeds its long‑term average, regardless of return‑volatility correlation. theoretical
Source sentence
In particular, under fast-scale dynamics, traders are advised to deleverage their portfolios when the current volatility level exceeds its long-term average, regardless of the return-volatility correlation.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 10 related papers). Coverage is partial, so this is neither replication nor refutation.
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.