Вы находитесь на странице: 1из 7

Valuation of Variance and Volatility Swaps

FINCAD Analytics Suite offers valuation of variance and volatility swapsboth with
model-independent replication strategies, and within the Heston Model. To download the
latest trial version of FINCAD Analytics Suite for free, contact a FINCAD Representative.
Overview
The classic derivatives that allow investors to take a view on volatility are straddles or
strangles. A long position in a straddle, for example, will generate a profit if the underlying
asset price moves up or down, or if the implied volatility rises. However, these options are
also sensitive to the underlying asset price, as the delta of a straddle or a strangle is zero
only when the option is at-the-money.
Unlike these options, variance and volatility swaps provide pure exposure to volatility. A
volatility swap is essentially a forward contract on future realized price volatility. At expiry
the holder of a long position in a volatility swap receives (or owes if negative) the
difference between the realized volatility and the initially chosen volatility strike, multiplied
by a notional principal amount. A variance swap is analogously a forward contract on future
realized price variance, which is the square of future realized volatility.
In both cases, at inception of the swap the strike is chosen such that the fair value of the
swap is zero. This strike is then referred to as fair volatility and fair variance, respectively.
Variance and volatility swaps can be used to speculate on future realized volatility, to trade
the spread between realized and implied volatility, or to hedge the volatility exposure of
other positions. For example, variance swaps can effectively protect against drops in the
underlying price: since volatility is generally negatively correlated with the underlying
price, a long position in a variance swap will generate a gain, if the underlying price drops
sharply.
Variance swaps can be replicated and valued in a model-independent manner, using a static
portfolio of European vanilla options [1]. For this reason, variance swaps are more popular
than volatility swaps - for which there exist only approximate static replication strategies.
On the other hand, both types of swaps may be valued within a calibrated Heston model [2].
Replication with Vanilla Options
The variance swap replication is accomplished using a portfolio of options with different
strikes. The construction of this portfolio can be understood intuitively in the Black Scholes
model.1 The sensitivity of a European option to the variance of the underlying asset price
depends on the asset price. This "variance vega"2 is largest when the underlying price is
closest to the strike of the option, and is also an increasing function of the strike. Both these
dependencies are illustrated in the graph below, which shows the variance vega as a
function of the underlying asset price for a range of strikes.
Figure 1: Variance vega for individual vanilla call option

The

variance vega of a portfolio of options that replicates the variance swap payoff must be
independent of the underlying price. To achieve this, each option has to be weighted by the
inverse of the strike squared. The following graph shows the variance vega of such a
portfolio: the sum of the above options, weighted appropriately. As can be seen, the
variance exposure of the portfolio is largely independent of the underlying asset price, as
long as the price lies within the range of option strikes.
Figure 2: Variance vega for a portfolio of vanilla options
In fact,
as the
spacing

between strikes is decreased and the range of strikes in the options portfolio increases, the
variance exposure becomes entirely independent of the underlying stock price. The rigorous
derivation given in [1] shows that such a portfolio indeed replicates the payoff of a variance
swap.
The FINCAD Analytics Suite function aaVarianceSwap2_port can be used to approximate a
variance swap with a finite number of options. From an input volatility smile table that lists
the available options, the function computes the value of each option and the number of
options at each strike required in the portfolio. On the other hand, the functions
aaVarianceSwap2_iv and aVarianceSwap2_p compute the fair variance and the fair value of

a variance swap, respectively.


The replication of a volatility swap is not as simple as that of a variance swap. Variance
arises naturally in the P&L of hedged options, which depends linearly on variance.
Volatility on the other hand should be viewed as a derivative of variance. As volatility is the
square root of variance, the relation is non-linear, which does not allow replication by a
static options portfolio. An approximate strategy replicates a volatility swap with unit
notional by a variance swap with notional of 1/(2.Kvol), where Kvol is the volatility strike of
the volatility swap.
FINCAD Analytics Suite provides the functions aaVolatilitySwap_port, aaVolatilitySwap_p
and aaVolatilitySwap_iv for calculating the approximate replicating portfolio, the fair value
and the implied volatility, respectively.
Valuation in the Heston Model
The replication strategies discussed above are largely model-independent, as they only
assume that the underlying asset price process be continuous. However, given some model
of the underlying asset price dynamics, variance and volatility swaps can also be priced. A
natural model to use is the Heston model of stochastic volatility. In this model the
underlying asset price S follows a standard lognormal process, and the variance V follows a
mean-reverting square root process:

where r is the risk-free interest rate, we have ignored dividends, and dW1 and dW2 are two
(correlated) standard Brownian motions. The five Heston model parameters are: the initial
variance V0, the long-term variance V, the speed of mean reversion K, the volatility of
volatility , and the correlation .
For the valuation of a variance swap, the expected total variance E[2] over the term of the
swap has to be calculated. In the Heston model this expectation value follows from the
differential equation for the variance process given above [2]:

Similarly, the expected total volatility required for the valuation of a volatility swap can be
computed in the Heston model. However, the expression is more complicated than the
variance expectation and its calculation involves a numerical integration [2].
FINCAD Analytics Suite provides the functions aaVarianceSwap_Heston_p,
aaVarianceSwap_Heston_iv, aaVolatilitySwap_Heston_p, and aaVolatilitySwap_Heston_iv
for the fair value calculation and for the calculation of the implied variance and volatility
for variance and volatility swaps respectively.
In addition, FINCAD Analytics Suite includes the functions
aaCalibrateVarianceSwaps_Heston and aaCalibrateVolatilitySwaps_Heston, which can be
used to calibrate the Heston model parameters to the term structure of variance and

volatility swaps respectively.


Example: Model-independent Valuation with a Replicating Portfolio
In this example we use the FINCAD Analytics Suite workbook "Variance or Volatility
Swap" to calculate the replicating portfolio for a variance swap and its fair variance which
would result in a zero fair value of the swap at inception. We also calculate the fair value of
the swap some time after inception. For the replication we have ten options that we enter in
the worksheet "Smile Table":

We enter the variance swap


contract on March 3, 2008
with expiration on May 2,
2008, a notional amount of
$100 per volatility point
squared, a variance strike of
0.04, corresponding to a
volatility of 20%, and an
accrual method of bus/252,
counting only market days.
The annually compounded
risk free rate is 3.5% and the underlying asset price is $100 with a dividend yield of 0%.
Using the input volatility smile table the replicating options portfolio is calculated on the
worksheet "Portfolio":

Hedging Portfolio
The first column in the "Hedging Portfolio" table lists the type of options required at the
strike price given in the second column. The implied volatility in the third column follows
directly from the given smile table. The fourth column is the option's value as computed in
the Black Scholes model. The replicating portfolio is given in the fifth column, which lists
the number of put and call options that are required at each strike. The final column is the
price of the position in each option. Clearly, the positions of the at-the-money options are
the most expensive. The total value of the portfolio is the sum of all option position prices
and amounts to about $40,000 for this variance swap.
We now calculate the fair variance and the associated risk statistics for the variance swap
using the worksheet "Fair Var". The result is
The fair
variance is
close to the
variance
strike. The
delta of the
variance swap
is very small;
it is different
from zero,
because the replication with just ten options is not perfect.
After two weeks we compute the present value of the variance swap. For simplicity we
assume that all inputs including the volatility smile are the same, except for the value date
which is now March 17. On the worksheet "Realized Var" we enter the realized prices for
the two week period and calculate the realized variance, which is 0.039 corresponding to a
volatility of 19.7%.
The realized
variance
(0.039) is
slightly
smaller than
the variance
strike (0.040)
of the swap.
The implied
variance on
the value date that follows from the options portfolio is 0.0395. The present value of the
swap is therefore slightly negative:

Example:
Valuation in the
Heston Model
We now evaluate
the fair variance
of the variance
swap in the
Heston model
with the function
aaVarianceSwap_Heston_iv. On the effective date (March 3), the Heston parameters
corresponding to the volatility smile given in the example above are an initial volatility of
19%, a long-term volatility of 25%, a speed of mean reversion of 1, a volatility of volatility
of 100%, and a correlation of -0.5. For the calculation we use the FINCAD Analytics Suite
workbook "Variance or Volatility Swap (Heston Model)". We enter the variance swap
details on the worksheet "Main" as shown in the following screenshot.
Note that the
valuation of
a variance
swap in the
Heston
model does
not depend
on the
volatility of
volatility or
on the
correlation.
The fair
variance of
the variance
swap
evaluates to
0.383 or
19.564% in
units of
volatility.
This value is
close to the
value we
calculated with the options portfolio replication strategy in the first example, as should be
expected. One would find exact agreement only for a perfectly calibrated model and an
options portfolio with continuous strikes.

Summary
FINCAD
Analytics Suite
provides
functions to
value variance
and volatility swaps using model-independent replication strategies as well as using the
Heston model. The former functions allow the user to compute portfolios that replicate the
floating leg of a variance swap. Based on this replication strategy, the fair variance and the
fair value including all risk statistics can be calculated. If the user has calibrated the Heston
model to European options or to variance or volatility swaps, the Heston model functions
can be used to compute the fair variance and the fair value of a variance or a volatility swap.
1 The mathematical derivation does not assume Black Scholes dynamics, but only requires
that the price process of the underlying asset be continuous, that is free of jumps.
2 Variance vega is the derivative of the option price with respect to variance, or equivalently
to the volatility squared.
References
[1] Demeterfi, K., Derman, E., Kamal, M. and Zou, J. (1999) 'A Guide to Volatility and
Variance Swaps', Journal of Derivatives, 6, 9-32; 'More Than You Ever Wanted to Know
About Volatility Swaps', Goldman Sachs Quantitative Strategies Research Notes, March 8,
1999.
[2] Gatheral, J. (2006) 'The Volatility Surface: A Practitioner's Guide', Wiley Finance.

Вам также может понравиться