Achieving Consistent Modeling Of VIX and Equities

Derivatives

Artur Sepp
Bank of America Merrill Lynch, London
[Link]@[Link]

Global Derivatives Trading & Risk Management 2012

Barcelona
April 17-19, 2012

1
Plan of the presentation

1) Discuss model complexity and calibration

2) Emphasize intuitive and robust calibration of sophisticated volatil-

ity models avoiding non-linear calibrations

3) Present local stochastic volatility models with jumps to achieve

joint calibration to VIX options and (short-term) S&P500 options

4) Present two factor stochastic volatility model to fit both the short-
term and long-term S&P500 option skews

2
References

Some theoretical and practical details for my presentation can be

found in:

1) Sepp, A. (2008) VIX Option Pricing in a Jump-Diffusion Model,

Risk Magazine April, 84-89
[Link]

2) Sepp, A. (2008) Pricing Options on Realized Variance in the He-

ston Model with Jumps in Returns and Volatility, Journal of Compu-
tational Finance, Vol. 11, No. 4, pp. 33-70
[Link]

3
Motivation. Model complexity and calibration

The conventional approach emphasizes the importance of ”closed-

form” solutions for a limited class of model chosen on the sole ground
of their analytical tractability

The justification is that since the model calibration is implemented by

a non-linear optimization, closed-form solution ”speed-up” the cali-
bration

I prefer the opposite route:

1) Develop robust PDE methods for generic one and two factor
stochastic volatility models

2) Obtain intuition about model parameters using approximation so

that non-linear fits can be avoided

4
Model specification I

Changes in the implied volatility surface can be factored as follows:

δσ(T, K) = β1 (change in short term ATM volatility)
+ β2 (change in term-structure of ATM volatility)
+ β3 (change in short term skew) (1)
+ β4 (change in term-structure of skew)
+ ...

Factors are reliable if:

1) They explain most of the variation
2) Can be estimated from historical and current data in intuitive ways
3) Can be explained in simple terms

5
Model specification II. Important factors for a volatility model
1) Short-term ATM volatility and term-structure of ATM volatility -
time-dependent level of the model ”ATM” volatility (the least of the
problem)

2) Short-term skew - jumps in the underlying and volatility factor(s)

3) Term-structure of skew and volatility of volatility - mean-reversion

and volatility of volatility of volatility factor(s)

The challenge is to re-produce these factors using stochastic volatil-

ity model with time-homogeneous parameters

In my presentation, I consider:

Part I - calibration of stochastic volatility with jumps to VIX skews

(SV model for short-term skew, up to 6m)

Part II - calibration of two-factor stochastic volatility model (SV

model for medium- and long-term skew, 3m-5y)
6
Part I. Joint calibration of SPX and VIX skews using jumps

I consider several volatility models to reproduce the volatility skew

observed in equity options on the S&P500 index:
Local volatility model (LV)
Jump-diffusion model (JD)
Stochastic volatility model (SV)
Local stochastic volatility model (LSV) with jumps

For each model, I analyze its implied skew for options on the VIX

I show that LV, JD and SV without jumps are not consistent with the
implied volatility skew observed in option on the VIX

I show that:
Only the SV model with appropriately chosen jumps can fit the im-
plied VIX skew

Importantly, that only the LSV model with jumps can fit both Equity
and VIX option skews
7
Motivation I
Find a dynamic model that can explain both the negative skew for
equity options on the S&P500 index (SPX) and positive skew for
options on the VIX

Implied volatility
135% of the SP500 and VIX

110%
S&P500, 1m
85% VIX, 1m

60%

35%

10%
50% 70% 90% 110% 130% 150% 170%
Strike %

Implied vols of 1m options on the SPX and the VIX as functions of

strike K% relative to 1m SPX forward and 1m VIX future, respectively
8
Motivation II. Dynamics
The VIX exhibits strong mean-reversion and jumps

0% 80%

-10% 70%

-20% 60%

ATM Vol, VIX

VIX
-30% ATM

Skew
50%
Skew
-40% 40%

-50% 30%

-60% 20%

-70% 10%

Aug-08

Aug-10
May-07

May-09

May-11
Jan-07

Dec-08

Dec-10
Oct-07

Oct-09

Oct-11
Mar-08

Mar-10
Page 1

Rights scale: Time series of the VIX and 1m ATM SPX implied
volatility (ATM)
Left scale: 1m 110% − 90% skew
9
The VIX I. Definition
The VIX is a measure of the implied volatility of SPX options with
maturity 30 days

Trading in VIX futures & options began in 2004 & 2006, respectively
Throughout 2004-2007, the average of the VIX is 14.66%
Throughout 2008-2011, the average of the VIX is 27.65%

Nowadays, VIX options are one of the most traded products on CBOE
with the average daily about 10% of all traded contracts

Formally, the VIX at time t, denoted by F (t), is the square root

of the expectation of the quadratic variance I(t, T ) of log-return
ln(S(T )/S(t))
s
1
F (t) = E [I(t, t + τT ) | S(t)] , τT = 30/365 (2)
τT
For a general non-linear pay-off function u:
" s ! #
1
U (t, T ) = E [u(F (T )) | S(t)] = E u E [I(T, T + τT ) | S(T )] | S(t)
τT
10
Valuation of the VIX options I. PDE method
Consider the augmented dynamics for S(t) and I(t):
dS(t) = σ(S(t))dW (t), S(0) = S
!2
σ(S(t)) (3)
dI(t) = dt, I(0) = 0
S(t)
1) Solve backward problem for V (t, S) = E [I(T, T + τT )] on time in-
terval [T, T + τT ]:
!2
σ(S)
Vt + LV = − , V (T + τT , S) = 0 (4)
S
where L is the backward operator:
1 2
L = σ (S)∂SS (5)
2
2) Solve backward problem for U (t, S) = E [u (F (T ))] on interval [0, T ]:
s !
1
Ut + LU = 0 , U (T, S) = u V (T, S) (6)
τT
The problem is similar to valuing an option on a coupon-bearing bond
11
Valuation of the VIX options II. Enhancement
Solve the forward problem for the density function of S, G(T, S 0), at
time T
e = 0 , G(0, S) = δ(S 0 − S)
GT − LG (7)
where Le is operator adjoint to L:
1 
2

LG
e = ∂
SS σ (S)G (8)
2
Then U (T, S) is priced by
Z ∞ s !
1
U (0, S) = u V (T, S 0) G(T, S 0)dS 0 (9)
0 τT
Generic implementation is done by means of finite-difference (FD)
methods that allow to tackle the problem for arbitrary choice of σ(S)

When using (9) we only need to solve 2 PDE-s, backward and forward,
to value VIX options with maturity T across different strikes

Stochastic volatility and jumps can be incorporated and treated by

FD methods (see Sepp (2011) for a review)
12
Specific Models I
Dupire LV and LSV models are ”black boxes” - for any admissible
set of model parameters, they guarantee calibration to equity skew

Instead, I consider simple model specifications to analyze their cali-

bration and implied VIX dynamics by fitting model parameters to:
1) at-the-money(ATM) implied volatility σATM, Eq
2) equity skew defined by:
SkewEq(α) ≡ σimp((1 + α)S) − σimp((1 − α)S)
where σimp(K) is SPX implied vol function of strike K and α = 10%

As an output, I consider VIX skew defined by:

SkewVIX(α) ≡ σimp,VIX((1 + α)S) − σimp,VIX(S),
where σimp(K) is VIX implied vol as function of strike K

Inputs on 31 October 2011 for 1m SPX and VIX options:

S(0) = 1253.3, σATM, Eq = 25%, SkewEq(10%) = −16%
F (0) = 30%, σATM,VIX = 101%, SkewVIX(10%) = 12%
13
Specific Models I. CEV model A: Basics
Consider the CEV process (Cox (1975)):
!β
S(t)
dS(t) = σ dW (t) (10)
S(0)
We can derive an approximation for implied vol at K = (1 − α)S:
!1−β 
S 1

σimp((1 − α)S) = σ 1 + α(1 − β) + O(α2)
S(0) 2
Thus
!1−β !1−β
S S
σATM,CEV(S) ≡ σ , SkewEq,CEV(α) ≡ −σ α(1 − β)
S(0) S(0)
Given σATM,Eq and SkewEq(α):
!1−β
S SkewEq(α)
σ = σATM , β =1+
S(0) ασATM,Eq

14
Specific Models II. CEV model C: Dynamics of the VIX
Approximate the VIX using:

σ(S(t)) σ (S(t))β
F (t) ≈ = = σ (S(t))β−1
S(t) S(t)
For the dynamics of F (t):
1
2
dF (t) = (β − 1)F (t)dW (t) + (β − 1)(β − 2)F 3(t)dt
2

Observations:
1), No mean-revertion

2), for β < 1, we have negative absolute correlation with the spot

15
Specific Models II. CEV model D: VIX implied volatility
We can show that approximately:
1
 
σimp,vix((1 − α)F ) = −(β − 1)F 1 − α + O(α2)
2
Thus:
1
σATM,VIX,CEV = (1 − β)F , SkewVIX,CEV = (1 − β)F α
2
Using implied parameters:
SkewEq(α) 1
σATM,VIX,CEV ≈ − , SkewVIX,CEV ≈ − SkewEq(α)
α 2
Both VIX ATM volatility and skew are proportional to the equity skew

Using SkewEq(10%) = −16% we get:

σATM,VIX,CEV = 160% (vs 100% actual)
SkewVIX,CEV = 8% (vs 12% actual)

CEV (and LV in general) tend to overestimate VIX ATM vol and

underestimate VIX skew
16
Specific Models II. CEV model E: Illustration

55% 250%
Implied SP500 1m Volatility Implied VIX 1m Volatility

45% 200%
Exact
Approximate
35% 150%
Market

25% 100% Model

Market

15% 50%

Strike % Strike %
5% 0%
75% 85% 95% 105% 115% 125% 70% 90% 110% 130% 150% 170%

Left: SPXskew; Right: the VIX skew

Implied model parameters: σ = 0.2474, β = −5.37

17
Specific Models III. Jump-diffusion A: Basics
Merton (1973) jump-diffusion with discrete jump in log-return ν :
dS(t) = σS(t)dW (t) + (eν − 1) S(t) [dN (t) − λdt] (11)
where N (t) is Poisson with intensity λ

We can show that to the leading order for small time-to-maturity:

λν
σimp(K) ≈ σ − ln (S/K )
σ

The short-term skew in JD is linear in jump size and its intensity

18
Specific Models III. Jump-diffusion B: Equity implied volatility
We have for small time-to-maturity:
λν
σimp(S) ≈ σ, SkewEq(α) ≈ 2α
σ
JD model is overdetermined so consider expected quadratic variance:
υ = σ 2 + λν 2
Introduce weight wjd, 0 < wjd < 1, as proportion of variance con-
2
tributed by jumps (a parameter for calibration) and take υ = σATM :
σ2 λν 2
1= 2 + 2 ≡ (1 − wjd) + wjd
σATM σATM
Thus, obtain equation to imply λ by:
λν 2 2
σATM
2
= wjd ⇒ λ = wjd
σATM ν2
Accordingly:
2αwjdσATM (SkewEq)2
ν= , λ= 2
, σ ≈ σATM
SkewEq(α) 4α wjd
19
Specific Models III. Jump-diffusion E: VIX dynamics
The quadratic variance in the jump-diffusion model is driven by:

dI(t) = σ 2dt + ν 2dN (t)

The variance swap

"Z #
1 T
V (t) = E I(t0)dt0 = σ 2 + λν 2
T 0
turns out to be a deterministic function

Thus, the model value of the VIX is constant and the implied volatility
of the VIX is zero - even though the JD model generates the equity
skew!

Thus property is shared by all space-homogeneous jump-diffusions

and Levy processes!

Nevertheless, jumps are needed for calibration of VIX skew

20
Specific Models III. Jump-diffusion F: Illustration

50% 140%
Implied SP500 1m Volatility Implied VIX 1m Volatility
120%
40%
Exact 100%
Approximate
Market 80% Model
30% Market
60%

40%
20%
20%
Strike % Strike %
10% 0%
75% 85% 95% 105% 115% 125% 70% 90% 110% 130% 150% 170%

Left: SPX skew; Right: VIX skew

Implied model parameters: wjd = 0.8, λ = 0.7762, ν = −0.2512,
σ = 0.1769

21
Specific Models IV. CEV with jumps A: Basics
Consider process:
!β
S(t)
dS(t) = σ dW (t) + (eν − 1) S(t) [dN (t) − λdt] (12)
S(0)
It turns out that the impact on skew from local volatility and jumps
is roughly linear and additive

Thus, specify the weight for the skew explained by jumps, wjd, and
CEV local volatility wlv , wlv = 1 − wjd

Using obtained results for CEV and JD models:

wlv SkewEq(α) 2ασATM wjd(SkewEq)2
β =1+ , ν= , λ=
ασATM SkewEq(α) 4α2
!1−β
S(t)
σ = σATM
S(0)

22
Specific Models IV. CEV with jumps B: The VIX
The variance swap can be approximated by:
 #2  !2β−2
1
Z T" 0
dS(t ) S(t)
V (0, T ) = E  dt0 ≈ σ 2 + λν 2,
T 0 S(t0) S(0)

The VIX is approximated using:

q
F (t) ≈ σ 2S 2β−2(t) + λν 2,

Observation: in CEVJD, β becomes smaller (in absolute terms) thus

the implied vol and skew of the VIX decrease

23
Specific Models V. CEV with jumps C: Illustration

50% 180%
Implied SP500 1m Volatility Implied VIX 1m Volatility
160%

40% 140%
Model 120%
Market
100%
30%
80%
60% Model
Market
20% 40%
20%
Strike % Strike %
10% 0%
75% 85% 95% 105% 115% 125% 70% 90% 110% 130% 150% 170%

Left: S&P500 skew; Right: the VIX skew

Implied model parameters: wjd = 0.4 (fitted to match the VIX ATM
volatility), λ = 0.2484, ν = −0.3140, σ = 0.2192, β = −3.3141

24
Specific Models V. Stochastic volatility A: Basics
Consider exponential OU stochastic volatility (Scott (1987) SV model,
Stein-Stein (1991) SV model is linear in Y (t)):

dS(t) = σS(t)eY (t)dW (0)(t)

(13)
dY (t) = −κY (t) + εdW (1)(t), Y (0) = 0
with dW (0)(t)dW (1)(t) = ρdt
κ is mean-reversion speed
ε is volatility-of-volatility

For equity skew, we can show approximately:

ρε 4σSkewEq(α)
σATM ≈ σ , SkewEq(α) ≈ α ⇒ρ=
4σ αε
For the VIX implied volatility, we can show approximately:
σε σε
σimp,vix = 2 , Skewvix(α) = − 2
α
F (t) 2F (t)
A) The ATM implied vol and the skew are proportional to σε;
B) The VIX skew is negative
25
Specific Models V. Stochastic volatility D: Illustration

50% 200% Implied VIX 1m Volatility

Implied SP500 1m Volatility
180%
160%
40%
140%
Model
Market 120%
30% 100%
80%
Model
60% Market
20%
40%
20%
Strike % Strike %
10% 0%
75% 85% 95% 105% 115% 125% 70% 90% 110% 130% 150% 170%

Left: SPX skew; Right: the VIX skew

Implied model parameters: κ = 4.48, ε = 2.69, ρ = −0.85, σ =
0.2194,

26
Specific Models. Summary
We have considered several models with conclusions:
Model Equity Skew VIX Skew Mean-revertion
CEV Yes Weak No
Jump-Diffusion Yes No No
CEV Jump-Diffusion Yes Yes (too steep) No
Stochastic Volatility Yes No Yes

To model the VIX skew, we need three components:

Local volatility (for equity and VIX skew)
Stochastic volatility (mean-reverting feature of the Vix)
Jumps in price and volatility (steeper equity and the VIX skew)

Two viable contenders:

1) SV model with jumps (space-inhomogeneous so some analytical
solutions possible - Sepp (2008))
2) LSV model with jumps, for example, CEV or non-parametric
(valuation by means of FD methods)

27
Specific Models VI. SV+Jump-Diffusion: Dynamics
Consider generic LSV model with jumps:

dS(t) = σS(t)eY (t)dW (0)(t) + S(t) [(eν − 1) dN (t) − λνdt]

dY (t) = −κY (t) + εdW (1)(t) + J v dN (t)
dI(t) = σ 2e2Y (t)dt + (J s)2 dN (t)
with dW (0)(t)dW (1)(t) = ρdt
N (t) is Poisson process with intensity λ
Jumps in S(t) and Y (t) are simultaneous and discrete with magni-
tudes ν < 0 and η > 0

Here:
σ = σ(t) - SV with jumps
S(t) β
 
σ = σ(t) S(0) - CEV+SV with jumps
σ = σ(loc,svj)(t, S(t)) - LSV with jumps (Lipton (2002))

28
Specific Models VI. SV+Jump-Diffusion: Illustration

50% 140% Implied VIX 1m Volatility

Implied SP500 1m Volatility
120%
40%
100%
Model
Market 80%
30%
60%
Model
40% Market
20%
20%
Strike % Strike %
10% 0%
75% 85% 95% 105% 115% 125% 70% 90% 110% 130% 150% 170%

Left: SPX skew; Right: the VIX skew

Implied model parameters: wsv = 0.65, wjd = 0.35 (fitted to match
the VIX ATM volatility), λ = 0.2173, ν = −0.314, ρ = −0.85, σ =
0.2227, κ = 4.48, ε = 1.7485 (= wsv ε),
b η = 1.30 (fitted to match to
fit the VIX skew)

29
Specific Models VI. CEV+SV+Jump-Diffusion: Illustration

50% 140% Implied VIX 1m Volatility

Implied SP500 1m Volatility
120%
40%
100%
Model
Market 80%
30%
60%
Model
40% Market
20%
20%
Strike % Strike %
10% 0%
75% 85% 95% 105% 115% 125% 70% 90% 110% 130% 150% 170%

Left: S&P500 skew; Right: the VIX skew

Implied model parameters: wjd = 0.35, wcev = 0.05, wsv = (1 −
wcev )(1 − wv ) = 0.6175 (fitted to match the VIX ATM volatility),
β = 0.77, λ = 0.2173, ν = −0.314, ρ = −0.85, σ = 0.2227, κ = 4.48,
ε = 1.6611 (= wsv ε),
b η = 1.30 (fitted to match to fit the VIX skew)

30
Specific Models VI. Non-parametric LSV+Jump-Diffusion

150% 120% 105%

1m 2m 3m
140% 100%
110%
130% 95%

100% 90%
120%
85%
110% 90%
80%
100%
80% Market, 2m 75% Market, 3m
90% Model, 2m Model, 3m
70%
70%
80% 65%
Market, 1m
K% Model, 1m
70% 60% 60%
70% 80% 90% 100% 110% 120% 130% 140% 150% 160% 170% 70% 80% 90% 100% 110% 120% 130% 140% 150% 160% 170%
60%
K% K%
70% 80% 90% 100% 110% 120% 130% 140% 150% 160% 170%

90% 85% 75%

4m 5m 6m
85% 80% 70%

75% 65%
80%
70% 60%
75%
65% 55%
70%
Market, 4m 60% Market, 5m 50% Market, 6m
65% Model, 4m Model, 5m Model, 6m
55% 45%
60%
50% 40%

55% K% 45% K% 35% K%

50% 40% 30%

70% 80% 90% 100% 110% 120% 130% 140% 150% 160% 170% 70% 80% 90% 100% 110% 120% 130% 140% 150% 160% 170% 70% 80% 90% 100% 110% 120% 130% 140% 150% 160% 170%

LSV model fit to VIX implied volatilities (the model is consistent with
the S&P500 implied volatilities by construction)
Implied model parameters: wjd = 0.35, wlv = 0.05, wsv = 0.6175,
λ = 0.2173, ν = −0.314, ρ = −0.80, κ = 4.48, ε = 1.6611, η = 1.30
31
Conclusions
For calibration of LSV model to SPX and VIX skews, a careful choice
between the stochastic volatility, local volatility and jumps is neces-
sary:

1) Local volatility: weight wlv - calibration parameter (typically

small)

2) Price jumps: weight wjd - calibration parameter (fitted to VIX

ATM volatility), λ and ν determined from given equity skew

3) Volatility jumps: η - calibration parameter (fitted to VIX skew)

4) SV parameters: SV weight wsv = 1 − wlv − wjd - calibration

variables, κ and ε from historical (implied) data

VIX options give extra information for calibration of LSV models

32
Part II. Joint calibration of medium- and long-term SPX skews
using two-factor stochastic volatility model

Observation: One-factor stochastic volatility model with time-independent

parameters can only calibrate to
A) either medium-term skews (up to 1 year) using large mean-reversion
and vol-of-vol
B) or long-term skews (from 1 year) using low mean-reversion and
vol-of-vol

Idea: introduce a two-factor stochastic volatility model with one fac-

tor for short-term skew and second factor for long-term skew

Challenge: find a proper weight (time-dependent) for the two factors

Bergomi (2005) assumes constant weights - I found it difficult to

calibrate his model with constant weights

33
Skew decay I
Consider 110%−90% market skew, Skew(Tn), for maturities {1m, 2m, ..., 60m}

Skew(Tn )
Compute skew decay rate: Rmarket(Tn) = Skew(T , n = 2, ..., 60
n−1 )

In addition, consider the following test functions:

Fα(Tn) = (Tn)−α
 −α
with decay rate Rα(Tn) = FF(T
α (Tn )
= Tn
Tn−1
α n−1 )

We observe:
1) for small T , the skew decay α ≈ 0.25

2) for large T , the skew decay α ≈ 0.50

Experiment - compute weighted average:

−α1 −α2
Fα1,α2 (Tn) = ω(Tn)Tn + (1 − ω(Tn))Tn , ω(t) = e−κt , κ = 2
Fα ,α (Tn )
with skew decay rate: Rα1,α2 (Tn) = F 1 (T 2
α1 ,α2 n−1 )
34
Skew decay II

0.99

0.97

0.95

Skew decay rate

Market rate skew
0.93
rate 1/T^0.5
0.91
rate 1/T^0.25

0.89 weighted sum

0.87
T
0.85
2m 6m 10m 14m 18m 22m 26m 30m 34m

For short maturities, the skew is proportional to T −0.25

For longer maturities, the skew is proportional to T −0.50

The decay of the market skew is reproduced using ”weighted” skew:

F0.25,0.50(Tn) = ω(Tn)Tn−0.25 + (1 − ω(Tn))Tn−0.5 , ω(t) = e−κt , κ = 2
35
Two-factor SV model I
Start with one-factor SV dynamics for S(t) and SV factor Y (t):

dS(t)/S(t) = µ(t)dt + σ exp {Y (t)} dW (0)

dY (t) = −κY (t)dt + εdW (1), dW (0)dW (1) = ρ
Then extend to two-factor SV dynamics :

dS(t)/S(t) = µ(t)dt + σ exp {ω(t)Y1(t) + (1 − ω(t))Y2(t)} dW (0)

dY1(t) = −κ1Y1(t)dt + ε1dW (1) , dY2(t) = −κ2Y2(t)dt + ε2dW (2)
dW (0)dW (1) = ρ01, dW (0)dW (2) = ρ02, dW (1)dW (2) = ρ12 ≡ ρ01ρ02
Here:
Y1(t) - SV factor for short-term volatility;
Y2(t) - SV factor for long-term volatility;
ω(t), 0 < ω(t) < 1, - weight between short-term and long-term factor:
1
 
ω(t) = exp − (κ1 + κ2) t
2
We expect: κ1 >> κ2, ε1 >> ε2, |ρ01| > |ρ02|

36
Two-factor SV model II. Calibration
The model is implemented using a 3-d PDE solver with a predictor-
corrector
Typically, we use N1 = 100 per year in time dimension, N2 = 200 in
spot dimension, N3 = N4 = 25 in volatility dimensions

For calibration, we solve the forward PDE to compute the density of

the underlying price and value European calls and puts at once (takes
about 1-minute to calibrate to ATM volatility up to 5 years)

To simplify the calibration we consider two quantities at given matu-

rity objects: the ATM volatility and 90% − 110% skew

1) calibrate two sets of parameters {κ, ε, ρ} for 1-factor SV model:

the first one - to short term skew (up to 1 year)
the second one - to long-term skew (from 1y to 5y)
2) Use these parameters for the 2-factor SV model with weight ω(t)

For all three models, piece-wise constant volatilities {σ(Tn)} are cal-
ibrated so that the model matches given market ATM volatility at
maturity times {Tn}
37
Two-factor SV model III. Illustration of calibration
Calibrated parameters for SPX (top) and STOXX 50 (bottom)
Short-term Long-term 2-f model
κ 3.50 0.65
ε 1.95 0.95
ρ -0.85 -0.80
ρ12 0.68
1 κ +κ 2.08
2( 1 2)

Short-term Long-term 2-f model

κ 2.40 0.55
ε 1.40 0.70
ρ -0.85 -0.80
ρ12 0.68
1 κ +κ 1.48
2( 1 2)

38
Two-factor SV model IV. Calibration to SPX
Term structure of skews for 1-f SV model calibrated up to 1y skews,
1-f SV model calibrated from 1y to 5y skews, 2-factor SV model

0% 0%
3m 9m 15m 21m 27m 33m 39m 45m 51m 57m 3m 9m 15m 21m 27m 33m 39m 45m 51m 57m

-2% -2%

-4% -4%

Market Skew
Market Skew
-6% Model Skew -6%
Model Skew

-8% -8%

-10% -10%

-12% -12%

0%
3m 9m 15m 21m 27m 33m 39m 45m 51m 57m

-2%

-4%

Market Skew
-6%
Model Skew

-8%

-10%

-12%

39
Two-factor SV model V. Calibration to STOXX 50
Term structure of skews for 1-f SV model calibrated up to 1y skews,
1-f SV model calibrated from 1y to 5y skews, 2-factor SV model

0% 0%
3m 9m 15m 21m 27m 33m 39m 45m 51m 57m 3m 9m 15m 21m 27m 33m 39m 45m 51m 57m
-1% -1%

-2% -2%

-3% -3%

-4% Market Skew -4%

Market Skew
Model Skew
-5% -5% Model Skew

-6% -6%

-7% -7%

-8% -8%

-9% -9%

0%
3m 9m 15m 21m 27m 33m 39m 45m 51m 57m
-1%

-2%

-3%

-4%
Market Skew

-5% Model Skew

-6%

-7%

-8%

-9%

40
Illustration VI. Model implied vol across strikes for SPX
A) 2-factor SV model with piecewise-constant volatility fits the term-
structure of market skews, unlike 1-factor SV model
B) Small discrepancies in model and market implied volatilities across
strikes are eliminated by introducing a parametric local volatility with
extra parameter to match the skew across strikes

45% 40% 40%

3m Market 3m 6m Market 6m 9m Market 9m
40% Model 3m 35% Model 6m 35% Model 9m

35%
30% 30%
Implied Vol

Implied Vol

Implied Vol
30%
25% 25%
25%
20% 20%
20%

15% 15% 15%

K% K% K%
10% 10% 10%
70% 75% 80% 85% 90% 95% 100% 105% 110% 70% 75% 80% 85% 90% 95% 100% 105% 110% 115% 70% 75% 80% 85% 90% 95% 100% 105% 110% 115%

35% 35% 35%

1y Market 12m 3y Market 36m 5y Market 60m
Model 12m Model 36m Model 60m
30% 30% 30%
Implied Vol

Implied Vol

Implied Vol
25% 25% 25%

20% 20% 20%

15% 15% 15%

K% K% K%
10% 10% 10%
70% 75% 80% 85% 90% 95% 100% 105% 110% 115% 70% 75% 80% 85% 90% 95% 100% 105% 110% 115% 70% 75% 80% 85% 90% 95% 100% 105% 110% 115%

41
Illustration VII. Model implied density for SPX

6.00

3m 6m 9m 12m 15m

5.00

18m 21m 24m 27m 30m

4.00

33m 36m 39m 42m 45m

Density

3.00

2.00 48m 51m 54m 57m 60m

1.00

0.00
0.00 0.18 0.37 0.55 0.73 0.91 1.10 1.28 1.46 1.65 1.83 2.01 2.20
Normalized Spot

Observe heavy tails for longer-maturities with concentration around

S = 0 (zero is assumed to be an absorbing barrier for my PDE solver)
42
Illustration VIII. The term structure of implied volatility-of-
volatility

180%

160%

140% 2-Factor SV

Implied VolVol
1-Factor SV
120%

100%

80%

60%
T
40%
3m 6m 1y 2y 3y 4y 5y

The term structure of implied volatility-of-volatility (implied log-normal

vol for ATM option on the realized variance) in 1- and 2-factor SV
model

43
Conclusions

I have presented:

1) Local stochastic volatility model with jumps for modeling of

short-term skews and illustrated its calibration to VIX options skews

I have shown that jumps are needed to fit the model to both SPX
and VIX skews

2) 2-factor stochastic volatility model for modeling of medium-

and long-dated skews

I have shown that only two-factor stochastic volatility model can fit
both medium- and long-dated skews

For both models, I have tried to emphasize an intuitive approach to

model calibration and avoid using ”blind” non-linear calibrations
44
Acknowledgment

During my work I benefited from interactions with:

Alex Lipton on his universal local stochastic volatility with jumps

(Lipton (2002))

Hassan El Hady, Charlie Wang and other members of BAML Global

Quantitative Analytics

The opinions and views expressed in this presentation are those of

the author alone and do not necessarily reflect the views and policies
of Bank of America Merrill Lynch

Thank you for your attention!

45
References
Bergomi, L. (2005), “Smile Dynamics 2”, Risk, October, 67-73
Cox, J. C. (1975), Notes on Option Pricing I: Costant Elasticity of
Variance Diffusions Unpublished Note, Stanford University
Lipton, A. (2002), “The vol smile problem”, Risk, February, 81-85
Merton, R. (1973), “Theory of Rational Option Pricing”, The Bell
Journal of Economics and Management Science 4(1), 141-183
Scott L. (1987) “ Option pricing when the variance changes ran-
domly: Theory, estimation and an application,” Journal of Financial
and Quantitative Analysis, 22, 419-438
Sepp, A. (2008) “Vix option pricing in a jump-diffusion model,” Risk,
April, 84-89
Sepp, A. (2011) “Efficient Numerical PDE Methods to Solve Cali-
bration and Pricing Problems in Local Stochastic Volatility Models”,
Presentation at ICBI Global Derivatives in Paris
Stein E., Stein J. (1991) “Stock Price Distributions with Stochastic
Volatility: An Analytic Approach”, Review of Financial Studies, 4,
727-752

46