Implied Volatility Expansions for VIX Options in Forward Variance Models
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Quantitative Finance. The day in quantitative finance: models of markets, risk and prices.
Ruth: We're looking at the paper: "Implied Volatility Expansions for VIX Options in Forward Variance Models".
Warren: Here's the gist of the paper. Implied volatility expansions for VIX options in forward variance models develop closed-form expansions for the implied volatility of VIX options within forward variance models to enable fast and accurate calibration without requiring numerical root-finding using option prices.
Ruth: Model Framework The work develops implied volatility expansions for VIX options within the class of forward variance models.
Warren: The VIX is designed to measure the 30-day implied volatility of a log-contract on the S&P 500 index.
Ruth: A natural way to model the joint dynamics of the S&P 500 spot price St and its implied volatility surface is to work directly with the term structure of implied variances of log-contracts, referred to as forward variances.
Paper summary: Ruth: So, looking at the title and the authors of this paper, Liao and Agarwal and Bourgey have presented a way to handle the implied volatility of VIX options using forward variance models by developing these explicit expansions. It's a method designed for speed and accuracy in calibration.
Warren: Essentially, they built formulas that let you compute implied volatility directly from market data without needing to solve those tough numerical problems that usually come up when calibrating these types of models.
Iris: The implication is that for people working with these specific variance structures, they can get very quick and reliable numbers for risk sensitivities on VIX derivatives, provided they stay within the conditions where the expansions hold true.
Ruth: That’s right. It’s about moving from a numerical inversion problem to a direct expansion problem, which is much more efficient for practical applications in markets where fast pricing is essential.
Conclusion: Ruth: So, we're wrapping up on this paper about these implied volatility expansions for VIX options in forward variance models by Liao and Agarwal and Bourgey.
Warren: They essentially developed explicit formulas to get the implied volatility of VIX options directly from forward variance models without needing to run those tough numerical root-finding procedures.
Iris: So, what does that mean practically? It means we can get really fast and accurate risk numbers for VIX derivatives using these models, provided we're within the right parameter settings.
Ruth: Exactly. They're moving away from slow numerical inversion towards a direct expansion approach, which is much faster for real-time calibration needs.
Warren: The speed gain is significant. The paper shows this expansion can be about one hundred thirty-one times faster than using the standard quadrature method in the standard Bergomi model for VIX maturities of one to six months.
Iris: That speed is impressive, but we have to keep an eye on the caveats. They do show that in mixed-kernel models—where you have a combination of two log-normal components—the Hermite-based expansion captures market VIX skews better, but only if those two proxy volatilities aren't too far apart.
Ruth: That’s the key limitation they point out. If those two underlying volatilities are too different, the accuracy drops in the mixed case.
Warren: And if you go into more extreme parameter regimes, even with that improvement, you might lose some accuracy. But it still gives you a reliable starting guess for those iterative root-finding procedures if things get weird.
Iris: It’s a strong result for calibration reliability then because they tested it against market data and found the calibrated smiles actually look quite good in the mixed rough Bergomi model.
Ruth: So, to put it simply, this work gives us a way to price VIX options more efficiently and reliably using forward variance models, but we gotta respect those parameter boundaries for the best results.
Warren: It’s a solid tool for quantitative traders looking to speed up their pricing processes when dealing with these specific volatility structures.
University of Glasgow · Western University · NYU
q-fin.CP, q-fin.MF
Submitted: 2026-04-28
Updated: 2026-11-06
Code: https://github.com/fbourgey/vix-expansion-bergomi
License: http://creativecommons.org/licenses/by/4.0/
Importance score: 82/100
The gist: Implied volatility expansions for VIX options in forward variance models develop closed-form expansions for the implied volatility of VIX options within forward variance models to enable fast and…
Key concepts
- Forward Variance Models
- These models describe the joint dynamics of the S&P 500 spot price and its implied volatility surface by focusing on the term structure of forward variances. The instantaneous forward variance curve is defined based on how this variance changes over time, providing a natural way to link spot prices and volatility dynamics.
- VIX
- The VIX is a measure designed to capture the 30-day implied volatility of a log-contract on the S&P 500 index. It serves as an important market indicator for gauging expected future volatility, which is central to modeling its options.
- Implied Volatility Expansion
- This is a direct formula derived for the VIX implied volatility that avoids numerical inversion of option prices. Instead of calculating the price and then solving for volatility, this expansion provides an explicit formula, making calibration much faster and more accurate.
Terminology
Summary
Implied volatility expansions for VIX options in forward variance models develop closed-form expansions for the implied volatility of VIX options within forward variance models to enable fast and accurate calibration without requiring numerical root-finding using option prices. p. 0 Our approach builds on weak-approximation techniques for
VIX option prices and yields explicit implied volatility expansions with computable correction terms. The resulting formulas enable fast and accurate calibration without requiring
numerical root-finding using option prices.Where the paper says it
Model Framework
The work develops implied volatility expansions for VIX options within the class of forward variance models. The VIX is designed to measure the 30-day implied volatility of a log-contract on the S&P 500 index. p. 0 Our implied volatility expansions are derived from VIX option price
approximations developed by [10]. The VIX was first introduced in 1993 by the Chicago Board Options Exchange (CBOE) and
is designed to measure the 30-day implied volatility of a log-contract on the S&P 500 index. In practice, the log-contract is replicated using market prices of listed vanilla options across
a continuum of strikes. Following [3], a natural way to model the joint dynamics of the S&P
500 spot price St and its implied volatility surface is to work directly with the term structure
of implied variances of log-contracts, referred to as forward variances. As is well known, these
forward variances are driftless under the risk–neutral measure and can be treated as additional
state variables alongside St
. The resulting models are referred to as forward variance models. The resulting models are referred to as forward variance models. In this framework, the instantaneous forward variance curve (ξ
u
t
)t≤u is defined by
ξ
u
t =
d
du
(u − t) ˆσt(u)
2
,
where σˆt(u)
2 denotes the implied variance of the log-contract with maturity u observed at time
t (equivalently, the variance swap volatility with maturity u at time t). In this work, we assume
that the process (ξ
u
t
)t≤u solves the following stochastic differential equation (SDE)
dξ
u
t = ξ
u
t Ku
(t) dWt
, (1)
where K is a deterministic L
2
-kernel function, and ξ
u
0
is a given initial forward variance curve. The solution to (1) is explicitly given by the log-normal process
ξ
u
t = ξ
u
0
exp
−
1
2
Z t
0
Ku
(s)
2 ds +
Z t
0
Ku
(s) dWs
,
∗University of Glasgow (2636546L@student. for all u ≥ t. Equivalently, the instantaneous log-forward variance Xu
t
:= ln(ξ
u
t
), with u ≥ t,
solves the SDE
dXu
t = −
1
2
Ku
(t)
2 dt + Ku
(t) dWt
.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Single-Kernel Expansions
The authors derive an explicit implied volatility expansion for the standard and rough Bergomi models building on the weak VIX option price approximation of [10]. This gives a direct implied volatility formula, rather than requiring numerical inversion of the price expansion. p. 1 First, building on the weak VIX option price approximation of
[10], recalled as Theorem 1, we derive an explicit implied volatility expansion for the standard
and rough Bergomi models; see Theorem 2 and Section 2. This gives a direct implied volatility
formula, rather than requiring numerical inversion of the price expansion. Under Assumptions 1, 2, and 3, the price of a VIX option with payoff φ(VIX2
T
)
admits the following approximation
E
φ(VIX2
T
)
= E
φ(VIX2
T,P )
+
X
3
i=1
γi ∂
i
ϵ E
φ
VIX2
T,P e
ϵ
ϵ=0 + Eφ, (12)
where the remainder term Eφ satisfies the bound |Eφ| ≤ C∆3(a1∧
a2
2
)
. The coefficients (γi)i∈{1,2,3}
are given explicitly by
γ1 :=
1
8∆ Z T +∆
T
ξ
u
0
E[VIX2
T
]
Z T
0
Ku
(t)
2 −
1
∆
Z T +∆
T
ξ
v
0
E[VIX2
T
]
Kv
(t)
2 dv
! Inspired by [7], we derive an approximation for the volatility implied by VIX call prices in
(14) by using relations between BS-Greeks and sensitivities of VIX option prices with respect
to log-spot x and volatility σ in (6). Under Assumptions 1, 2 and 3, the VIX implied volatility σBS(k, T) defined in (7)
is approximated by
σBS(k, T) ≈
σeP
2
+
γ2
2σeP T
+
3γ3
8σeP T
−
γ3
σe
3
P
T2
(xP − k), (17)
where σeP = √σP
T
, xP =
µP
2 +
σ
2
P
8
, and the coefficients (γi)i∈{1,2,3} are given inWhere the paper says it
Where the paper says it
Where the paper says it
Mixed-Kernel Extensions
The construction is extended to mixed-kernel forward variance models where the forward variance process (ξu t)t≤u is modeled as a convex mixture of two log-normal components. The VIX option price approximation in (23) is given by E phi(VIX2 T) = E phi(VIX2 T,P) + X 2 j=1 X 3 i=1 γi,j Pi,j + Ephi where the remainder term satisfies Ephi ≤ C ∆3(a1∧a2/2). p. 4 Under Assumptions 1, 2, and 3, the price of a VIX option with payoff φ(VIX2
T
)
admits the following approximation
E
φ(VIX2
T
)
= E
φ(VIX2
T,P )
+
X
3
i=1
γi ∂
i
ϵ E
φ
VIX2
T,P e
ϵ
ϵ=0 + Eφ, (12)
where the remainder term Eφ satisfies the bound |Eφ| ≤ C∆3(a1∧
a2
2
)
. The coefficients (γi)i∈{1,2,3}
are given explicitly by
γ1 :=
1
8∆ Z T +∆
T
ξ
u
0
E[VIX2
T
]
Z T
0
Ku
(t)
2 −
1
∆
Z T +∆
T
ξ
v
0
E[VIX2
T
]
Kv
(t)
2 dv
! Inspired by [7], we derive an approximation for the volatility implied by VIX call prices in
(14) by using relations between BS-Greeks and sensitivities of VIX option prices with respect
to log-spot x and volatility σ in (6). Under Assumptions 1, 2 and 3, the VIX implied volatility σBS(k, T) defined in (7)
is approximated by
σBS(k, T) ≈
σeP
2
+
γ2
2σeP T
+
3γ3
8σeP T
−
γ3
σe
3
P
T2
(xP − k), (17)
where σeP = √σP
T
, xP =
µP
2 +
σ
2
P
8
, and the coefficients (γi)i∈{1,2,3} are given inWhere the paper says it
Where the paper says it
Numerical Validation and Calibration
Numerical tests show that the expansions perform well with relative errors below 1% in both standard and rough Bergomi models for VIX maturities T ∈ 1, 3, and 6 months. The implied volatility expansion is approximately 131× faster than the quadrature method in the standard Bergomi model. p. 6 37 µs (Expansion). Thus, the implied volatility expansion is approximately 131× faster than the quadrature method
and 32× faster than the weak approximation in the standard Bergomi model, and approximately
52,750× faster than the Monte Carlo method and 27× faster than the weak approximation in
the rough Bergomi model. We illustrate the performance of the proposed
expansions in both standard and rough Bergomi-type models, as well as in mixed specifications, and demonstrate their accuracy through numerical experiments. 1 Introduction
Accurate closed-form implied volatility expansions allow industry practitioners to efficiently
calibrate their models to market data. Therefore,
CP ≈ C
BSxP , k, σeP
2 + ∆σ
+ δ ∂xC
BSxP , k, σeP
2
. 3 Numerical tests
In this section, we evaluate the accuracy of the VIX implied volatility expansion (17) (referred
to as Expansion) under the standard and rough Bergomi models, and compare it with implied
volatilities obtained from the weak approximation in Theorem 1 (referred to as Approx ).Where the paper says it
Where the paper says it
Where the paper says it
Conclusion and Limitations
In the mixed-kernel case, the Hermite-based expansion captures market VIX skews more effectively, provided the two proxy volatilities are not too far apart. In more extreme parameter regimes, the expansion may lose accuracy but still provides reliable initial guesses for standard root-finding procedures. p. 1 3, together with the calibration study
in Section 4, show that the expansions allow rapid calibration across a wide range of parameters
and that mixed specifications capture market VIX skews more effectively than single log-normal
kernels, a phenomenon consistent with the limitations documented in [14]. The accuracy of
our formulas deteriorates in mixed-kernel models only in extreme parameter regimes, where
the formulas still provide reliable initial guesses which can be used for standard root-finding
procedures. In the
mixed rough Bergomi model, the main term reads
E
φ(VIX2
T,P )
= E
φ
X
2
j=1
λj e
µP,j+σP,j Z
, Z ∼ N (0, 1), (24)
and the correction terms are given, for i ∈ {1, 2, 3} and j ∈ {1, 2}, by
Pi,j = E
h
Ψ
(i−1)
j
(Xj )
i
, Xj := µP,j − L + σP,j Z, L := lnE[VIX2
T
]
, (25)
where
Ψj (x) = λj e
x E[VIX2
T
] φ
′
E[VIX2
T
] Sj (x)
,
and
S1(x) := λ1 e
x + λ2 e
(
η1
η2
−1)(L−µP,2)+ η2
η1
x
, (26)
S2(x) := λ1 e
(
η2
η1
−1)(L−µP,1)+ η1
η2
x + λ2 e
x
. (27)
The formulas for the mixed standard Bergomi model are obtained by replacing the kernel and ηj
with ωj in (26) and (27).Where the paper says it
Where the paper says it
Additional Numerical Results
The numerical results under the mixed standard Bergomi model show that the performance of (45) is slightly worse than that of (40) in Scenario 1 where ω1 and ω2 are close. The implied volatility expansion (45) generally performs worse than the expansion proposed in Theorem 4. p. 6 02, from [9]
1
. In all four cases, the implied volatility
expansion performs well with relative errors below 1%. 1 Our contributions
Our contributions are threefold. First, building on the weak VIX option price approximation of
[10], recalled as Theorem 1, we derive an explicit implied volatility expansion for the standard
and rough Bergomi models; see Theorem 2 and Section 2.Where the paper says it
Where the paper says it
Additional Calibration Tests
The calibration results obtained from our implied volatility expansion can be regarded as reliable when compared to those from the weak price approximation. The calibrated smiles reported in Figure 10 show that the implied volatility expansion formula (40) captures the market smiles quite well. p. 7 4
log(K/F)
80
100
120
140
160
180
VIX implied volatility (in %)
1M
3M
6M
Approx
Expansion
0. 4
log(K/F)
55
50
45
40
35
30
25
20
Relative error (in bps)
1M
3M
6M
Approx
Expansion
Figure 1: VIX smiles in the standard Bergomi model. 3 Numerical tests
In this section, we evaluate the accuracy of the VIX implied volatility expansion (17) (referred
to as Expansion) under the standard and rough Bergomi models, and compare it with implied
volatilities obtained from the weak approximation in Theorem 1 (referred to as Approx ). In the
standard Bergomi model, implied volatilities are computed via two-dimensional Gauss-Legendre
quadratures (quadrature) with 120 nodes each.Where the paper says it
Where the paper says it
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Improved system can perform fast and accurate calibration without requiring numerical root-finding using option prices by developing closed-form expansions for implied volatility within forward variance models.
-
Improved system can derive explicit implied volatility formulas directly from market data approximations, replacing the need for numerical inversion of price expansions in standard and rough Bergomi models.
-
Improved system can extend these derivations to mixed-kernel forward variance models, allowing it to capture market VIX skews more effectively than single-kernel models, provided the two proxy volatilities are not too far apart.
-
Improved system can perform rapid calibration across a wide range of parameters by utilizing the derived expansions for standard and mixed-kernel forward variance frameworks.
-
Improved system can be used for risk sensitivity calculations based on current market conditions by efficiently computing risk sensitivities of derivative trading books using the developed implied volatility expansions.
Abstract
We develop expansions for the implied volatility of VIX options within the class of forward variance models based on weak-approximation techniques for VIX option prices. In single-kernel models, such as the standard and rough Bergomi models, the expansion is fully explicit, and we prove that its error is of the same order as that of the underlying price approximation. In mixed-kernel models, we propose two expansions. The first one is fully explicit and relies on a Hermite expansion of the price, which we rewrite exactly as a Black-Scholes price with a shifted volatility, up to explicit residual terms. The second one uses the implied volatility of a mixed log-normal proxy, computed numerically, as its zeroth-order term, and we prove that it achieves the same order of accuracy as in the single-kernel case. The resulting formulas yield implied volatilities directly, rather than by inverting approximate option prices.
Related papers
- AFNO: Universal Certifiably-Arbitrage-Free Operator Learning Framework for European Call-Price Surfaces
- Diffusion models for dynamic volatility surface generation and data-driven hedging
- Rethinking Synthetic Scenario Realism: Compatibility, Not Fidelity, Drives Hedging Performance
- CosmosAlpha: Multi-Relational Graph Search with Pareto-Guided Exploration for LLM-Driven Alpha Mining
- Diversification Across Predictors: Heterogeneous Ensemble Learning for Cross-Sectional Return Prediction
- DSINC beyond Heston: the 3/2 and other non-affine stochastic-volatility models