Paper: arXiv 2504.14343
Authors: Christoph Reisinger, Maria Olympia Tsianni
Abstract
We analyse a Monte Carlo particle method for the simulation of the calibrated Heston-type local stochastic volatility (H-LSV) model. The common application of a kernel estimator for a conditional expectation in the calibration condition results in a McKean-Vlasov (MV) stochastic differential equation (SDE) with non-standard coefficients. The primary challenges lie in certain mean-field terms in the drift and diffusion coefficients and the \(1/2\)-Hölder regularity of the diffusion coefficient. We establish the well-posedness of this equation for a fixed but arbitrarily small bandwidth of the kernel estimator. Moreover, we prove a strong propagation of chaos result, ensuring convergence of the particle system under a condition on the Feller ratio and up to a critical time. For the numerical simulation, we employ an Euler-Maruyama scheme for the log-spot process and a full truncation Euler scheme for the CIR volatility process. Under certain conditions on the inputs and the Feller ratio, we prove strong convergence of the Euler-Maruyama scheme with rate \(1/2\) in time, up to a logarithmic factor. Numerical experiments illustrate the convergence of the discretisation scheme and validate the propagation of chaos in practice.
Complexity vs Empirical Score
- Math Complexity: 9.5/10
- Empirical Rigor: 3.0/10
- Quadrant: Lab Rats — high math complexity, limited empirical testing
Why this score: The paper is highly theoretical, proving existence, uniqueness, and propagation of chaos for a McKean-Vlasov SDE with irregular coefficients using advanced stochastic analysis, yet it only presents minimal numerical experiments to illustrate the convergence of the scheme, lacking backtests or practical trading implementation.
Research Flowchart
flowchart TD A["<b>Research Goal:</b><br>Analyse Monte Carlo particle method<br>for calibrated H-LSV model"] --> B["<b>Key Methodology:</b><br>1. McKean-Vlasov SDE Analysis<br>2. Propagation of Chaos<br>3. Euler-Maruyama Scheme"] B --> C["<b>Mathematical Framework:</b><br>Conditional Expectation via Kernel Estimator<br>Mean-field terms in SDE coefficients"] C --> D["<b>Computational Process:</b><br>Particle System Simulation<br>Log-spot (Euler-Maruyama)<br>CIR Volatility (Full Truncation)<br>Condition: Feller Ratio"] D --> E["<b>Key Findings/Outcomes:</b><br>1. Well-posedness for fixed bandwidth<br>2. Strong Propagation of Chaos (up to critical time)<br>3. Strong convergence rate 1/2 (log factor)<br>4. Numerical validation of convergence"]
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 Monte Carlo particle method for simulating the calibrated Heston-type local stochastic volatility (H‑LSV) model, including a strong propagation of chaos result and a strong convergence analysis of an Euler–Maruyama scheme for the log‑spot process.
Source sentence
We analyse a Monte Carlo particle method for the simulation of the calibrated Heston-type local stochastic volatility (H-LSV) model. In our work, to approximate the conditional expectation, we use the particle method introduced for a general LSV model by Guyon and Henry-Labord`ere in [27].
Assumes 4
- (A1) The function (t,x)↦σDup(t,ex) is bounded, Lipschitz in x and 1/2‑Hölder in t. model
Source sentence
We assume that (A1) (t, x) →σDup(t, ex) is bounded, Lipschitz in x, and 1 2-H¨older in t, so that there exist positive constants A1, LDup such that for all x1, x2 ∈R, and t1, t2 ∈[0, T], |σDup(t1, ex1)| ≤A1, and σDup(t1, ex1) −σDup(t2, ex2) ≤LDup �|t1 −t2|1/2 + |x1 −x2| .
- (A2) The kernel K(·) is bounded and Lipschitz continuous. model
Source sentence
(A2) K(·) is bounded and Lipschitz continuous so that there exist positive constants A2, LK such that for all x1, x2 ∈R, |K(x1)| ≤A2, and |K(x1) −K(x2)| ≤LK|x1 −x2|.
- The Feller ratio parameter ν satisfies ν ≥ 1. model
Source sentence
Let Assumptions 1 hold and also assume that ν ≥1 and E[|X0|4] < ∞.
- The initial log‑spot X0 has finite fourth moment, E[|X0|^4] < ∞. model
Source sentence
Let Assumptions 1 hold and also assume that ν ≥1 and E[|X0|4] < ∞.
Data
- Synthetic market implied volatility surface synthetic unknown
Source sentence
Instead of using real market data, we simulate a “market” implied volatility surface using pre– calibrated Heston parameters, V0 = 0.0094, θ = 0.0137, κ = 1.4124, ρ = −0.1194, and ξ = 0.2988, sourced from [12].
Costs
Not applicable
Finds 5
- The regularised MV‑SDE (6) admits a unique strong solution up to a critical time T*. theoretical
Source sentence
Then, there exists a unique strong solution to equation (6) up to a critical time T ∗, for T ∗as in (11).
- A strong propagation of chaos result is established, guaranteeing convergence of the particle system to the MV‑SDE under a condition on the Feller ratio. theoretical
Source sentence
Moreover, we prove a strong propagation of chaos result, ensuring convergence of the particle system under a condition on the Feller ratio and up to a critical time. We are now well-equipped to prove a strong propagation of chaos result that establishes the convergence of the particle system (8) to the McKean–Vlasov SDE (10), and in the following sec-tion a timestepping approximation scheme.
- The Euler–Maruyama scheme for the log‑spot process converges strongly with rate 1/2 in time (up to a logarithmic factor) under conditions on the inputs and the Feller ratio. theoretical
Source sentence
Under certain conditions on the inputs and the Feller ratio, we prove strong convergence of the Euler-Maruyama scheme with rate $1/2$ in time, up to a logarithmic factor.
- Numerical experiments show strong convergence rates of 0.43, 0.51 and 0.51 for different κ values, and 0.51 for ν=3.00, confirming the theoretical predictions. performance
Source sentence
We observe strong convergence with rates 0.43, 0.51, and 0.51, as shown in Figures 1, 2, and 3, respectively. For ν1 = 0.33 < ν∗the rate is slower, as expected, while for ν3 = 4.00 > ν∗we observe half-order convergence as suggested by the theory.
- Strong propagation of chaos is observed empirically with an RMSE convergence rate of approximately 0.40 as the number of particles increases. performance figure from elsewhere in the paper
Source sentence
Specifically, we plot in log-log scale the root-mean-square error (RMSE) for increasing values of the number of particles N. We measure error := v u u t 1 2N 2N X i=1 �Si,2N T −eSi,2N T 2 , (51) where both systems {Si,2N T }i=1,...,2N and {eSi,2N T }i=1,...,2N are driven by the same Brownian mo-tion. However, for the particles denoted by eSi,2N T , the leverage function is determined only using the first N
Checked by others
Supported by
- Strong existence and uniqueness of a calibrated local stochastic volatility model
Compare the findings
This paper: Strong propagation of chaos is observed empirically with an RMSE convergence rate of approximately 0.40 as the number of particles increases.
That paper: Propagation of chaos holds: the empirical particle system converges to the McKean‑Vlasov limit as the number of particles M→∞.
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.