Analytical approximation of the transition density in a local volatility model (Q432231): Difference between revisions
From MaRDI portal
Latest revision as of 03:43, 14 July 2024
scientific article; zbMATH DE number 6751103
- The exact Taylor formula of the implied volatility
Language | Label | Description | Also known as |
---|---|---|---|
English | Analytical approximation of the transition density in a local volatility model |
scientific article; zbMATH DE number 6751103 |
|
Statements
Analytical approximation of the transition density in a local volatility model (English)
0 references
The exact Taylor formula of the implied volatility (English)
0 references
3 July 2012
0 references
21 July 2017
0 references
The paper is mainly concerned with one-dimensional local volatility market models of the form \[ \text{d}S_t=\mu(t, S_t)\text{d} t+\sigma(t, S_t)S_t\text{d}W_t. \] Exploiting the analytical approximation of function \(\sigma\) the authors obtain an expansion of the transition density \(\Gamma(t, S_0; T, S)\) of \(S\) of the form \[ \Gamma(t, S_0; T, S)\approx \Gamma^0(t, S_0; T, S)+\sum_{n=1}^NJ_{S_0}^n\Gamma^0(t, S_0; T, S), \quad N\geq1, \] where the main term \(\Gamma^0(t, S_0; T, S)\) is the density in a suitable Black-Scholes model, while \(J_{S_0}^n\) is a differential operator containing derivatives w.r.t. \(S_0\). This is the expansion in terms of BS-Greeks. An iterative algorithm for the explicit expressions of the operators \(J_{S_0}^n\) is constructed. This algorithm is straightforward to implement by using a symbolic computation software. The analytical approximation formulae are tested with a Monte-Carlo simulation in the context of one-dimensional local volatility models (standard quadratic and CEV).
0 references
In this work, a model driven by a multidimensional local diffusion (a Feller process which is also an inhomogeneous local diffusion) is studied, and an exact Taylor formula for implied volatility \(\sigma=\sigma(t,x; T,k)\) related to a call option with log-strike \(k\), maturity \(T\) and log-price \(x\) of the underlying asset at time \(t\) is proved. The obtained exact Taylor formula holds in a parabolic region of \((T,k)\) close to expiry and at-the-money, namely \(|x-k|\leq \lambda\sqrt{T-t}\) for an arbitrary \(\lambda>0\). The results are proved under mild conditions and apply to many popular models, including CEV, Heston and SABR, among others. Also, one can use them to study asymptotic behaviour of IV generated by VIX options where vertical limits (with \(T\to t+\) and \(k\) fixed) known in the literature are not sufficient. The main idea is to obtain sharp bounds on the difference between \(\sigma\) (resp. its derivatives \(\partial^q_T\partial^m_k \sigma\)) and a fully explicit approximation \(\bar{\sigma}\) (resp. its derivatives \(\partial^q_T\partial^m_k \bar{\sigma}\)) as \((T,k)\to(t,x)\) within the considered region. Here \(2q+m\leq N\) and \(N\) is the order of approximation. The approximation \(\bar{\sigma}\) was introduced and studied in an earlier paper by \textit{M. Lorig} et al. [Math. Finance 27, No. 3, 926--960 (2017; Zbl 1422.91713)].
0 references
local volatility
0 references
analytical approximation
0 references
heat kernel expansion
0 references
Black-Scholes formula
0 references
transition density
0 references
implied volatility
0 references
local-stochastic volatility
0 references
local diffusions
0 references
Feller process
0 references
0 references
0 references
0 references
0 references
0 references
0 references
0 references
0 references
0 references