You are on page 1of 118

Pricing and hedging of FX plain vanilla options

An empirical study on the hedging performance


of a dynamic Black-Scholes delta hedge with updating implied volatility
under the assumption of Heston and Black-Scholes underlying dynamics, respectively,
in the interpolation/extrapolation of option prices.
Jannik Nrgaard
MSc Finance Thesis
Supervisor:
Elisa Nicolato
Department of Business Studies
Aarhus School of Business,
University of Aarhus
August 2011
c Jannik Nrgaard 2011
The thesis has been typed with Computer Modern 12pt
Layout and typography is made by the author using L
A
T
E
X
The author wish to thank the following: My supervisor Elisa Nicolato, Researcher at
Aarhus School of Business in the Finance Research Group, Aarhus, Denmark for advice.
A thanks to Matthias Thul, PhD Candidate in Finance at Australian School of Business,
New South Wales, Sydney, Australia for answering questions. I thank the people whove
helped me gain access to the Bloomberg terminals at the University of Aarhus as well as
the employees at the Bloomberg service desk for answering my questions. Lastly, thanks
to Nordea for providing me the access to the Nordea Markets platform, Nordea Analytics,
from where Ive gathered supplementary data.
I want to take the opportunity to thank my parents
for their unconditional support during my years of study.
Abstract
The thesis shows evidence against the Black-Scholes assumption of a diu-
sion process for the log asset price that has stationary and independent normal
increments resulting in a log-normal distribution of asset returns by consider-
ing a time-series of spot rates on the EURUSD and the USDJPY covering a
period of recent years. Observations of distributions exhibiting high peakness
and "fat tails" as well as observations of volatility clustering are supported by
empirical evidence of heteroscedasticity, implying that the volatility of returns
is not constant over time, and evidence of autocorrelation.
In order to calibrate The Heston model and the Black-Scholes model to
market prices on plain vanilla call options the thesis deals with the foreign ex-
change specic quoting conventions and considers the dierence here between
the EURUSD and the USDJPY. A data set of 371 recent trading days are
collected from published quotes on Bloomberg where each model is calibrated
to a set of option prices on each day to obtain an overall goodness of t mea-
sure that shows the superior performance of the Heston model. In the case
of both underlying FX pairs the volatility surface is negatively skew shaped
throughout the period considered.
Based on the calibrations a large scale hedging experiment is set up where
a number of plain vanilla call options with dierent maturities and strikes is
sold on each day. A dynamic BS Delta hedge with updating implied volatility
simulated in each of the models results in a better hedging performance when
the underlying dynamics follows the Heston model. Furthermore we observe
that the hedging error is correlated with the underlying returns.
Contents
Contents i
List of Figures iii
List of Tables v
1 Introduction 1
2 Problem Statement 4
2.1 Research Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Delimitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3 The FX Market 8
3.1 FX rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.2 FX forward contract . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.3 FX options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
4 The Black-Scholes model 12
4.1 Geometric Brownian Motion . . . . . . . . . . . . . . . . . . . . . . . 12
4.2 The Black-Scholes equation . . . . . . . . . . . . . . . . . . . . . . . 13
4.3 The Garman-Kohlhagen formula . . . . . . . . . . . . . . . . . . . . . 15
4.4 Simulation of the Black-Scholes model . . . . . . . . . . . . . . . . . 18
5 Empirical facts 19
5.1 The distribution of FX returns . . . . . . . . . . . . . . . . . . . . . . 19
6 The Heston model 24
6.1 The process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
6.2 The solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
6.3 Simulation of the Heston model . . . . . . . . . . . . . . . . . . . . . 27
7 Market data 29
i
7.1 Quoting conventions . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
7.2 Retrieving the implied volatility . . . . . . . . . . . . . . . . . . . . . 31
8 Data description 35
9 Calibration of the models 36
9.1 Building the market implied volatility surface . . . . . . . . . . . . . 36
9.2 Calibration of the Heston model . . . . . . . . . . . . . . . . . . . . . 37
9.3 Calibration of the Black-Scholes Model . . . . . . . . . . . . . . . . . 38
9.4 Objective Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
9.5 Calibration results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
10 Empirical study on the hedging performances 49
10.1 Size of the study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
10.2 Strike levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
10.3 The hedging portfolio . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
10.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
11 Conclusion 55
Bibliography 57
A Retrieving the strike price corresponding to a premium included
Delta 60
B Building the market implied volatility surface 63
C Calibration of the Heston model 76
D Calibration of the Black-Scholes model 82
E Simulation of the Heston model 85
F Simulation of the Black-Scholes model 89
G No hedge 92
H Dynamic BS Delta Hedge with updating imp. vol. from the
Heston model 97
I Dynamic BS Delta Hedge with updating imp. vol. from the
Black-Scholes model 109
ii
List of Figures
5.1 Empirical sample frequency for EURUSD . . . . . . . . . . . . . . . . . . 20
5.2 Empirical sample frequency for USDJPY . . . . . . . . . . . . . . . . . . 20
5.3 Q-Q plot for EURUSD . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
5.4 Q-Q plot for USDJPY . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
5.5 Daily log returns for EURUSD . . . . . . . . . . . . . . . . . . . . . . . 20
5.6 Daily log returns for USDJPY . . . . . . . . . . . . . . . . . . . . . . . . 20
5.7 Autocorrelation for EURUSD . . . . . . . . . . . . . . . . . . . . . . . . 20
5.8 Autocorrelation for USDJPY . . . . . . . . . . . . . . . . . . . . . . . . 20
5.9 Rolling historic volatility for EURUSD . . . . . . . . . . . . . . . . . . . 21
5.10 Rolling historic volatility for USDJPY . . . . . . . . . . . . . . . . . . . 21
9.1 One week moving average of . . . . . . . . . . . . . . . . . . . . . . . . 43
9.2 One week moving average of . . . . . . . . . . . . . . . . . . . . . . . . 43
9.3 One week moving average of . . . . . . . . . . . . . . . . . . . . . . . . 43
9.4 One week moving average of . . . . . . . . . . . . . . . . . . . . . . . . 43
9.5 One week moving average of v
t
. . . . . . . . . . . . . . . . . . . . . . . 43
9.6 Call prices 1M on EURUSD 1/4/2010 . . . . . . . . . . . . . . . . . . . 45
9.7 Call prices 1Y on EURUSD 1/4/2010 . . . . . . . . . . . . . . . . . . . . 45
9.8 Imp. vol. 1M on EURUSD 1/4/2010 . . . . . . . . . . . . . . . . . . . . 45
9.9 Imp. vol. 1Y on EURUSD 1/4/2010 . . . . . . . . . . . . . . . . . . . . 45
9.10 Call prices 1M on EURUSD 6/1/2010 . . . . . . . . . . . . . . . . . . . 45
9.11 Call prices 1Y on EURUSD 6/1/2010 . . . . . . . . . . . . . . . . . . . . 45
9.12 Imp. vol. 1M on EURUSD 6/1/2010 . . . . . . . . . . . . . . . . . . . . 45
9.13 Imp. vol. 1Y on EURUSD 6/1/2010 . . . . . . . . . . . . . . . . . . . . 45
9.14 Call prices 1M on USDJPY 1/4/2010 . . . . . . . . . . . . . . . . . . . . 46
9.15 Call prices 1Y on USDJPY 1/4/2010 . . . . . . . . . . . . . . . . . . . . 46
9.16 Imp. vol. 1M on USDJPY 1/4/2010 . . . . . . . . . . . . . . . . . . . . 46
9.17 Imp. vol. 1Y on USDJPY 1/4/2010 . . . . . . . . . . . . . . . . . . . . . 46
9.18 Call prices 1M on USDJPY 6/1/2010 . . . . . . . . . . . . . . . . . . . . 46
9.19 Call prices 1Y on USDJPY 6/1/2010 . . . . . . . . . . . . . . . . . . . . 46
iii
9.20 Imp. vol. 1M on USDJPY 6/1/2010 . . . . . . . . . . . . . . . . . . . . 46
9.21 Imp. vol. 1Y on USDJPY 6/1/2010 . . . . . . . . . . . . . . . . . . . . . 46
10.1 Development in EURUSD spot rate . . . . . . . . . . . . . . . . . . . . . 53
10.2 Development in USDJPY spot rate . . . . . . . . . . . . . . . . . . . . . 53
iv
List of Tables
5.1 Jarque-Bera test on normality . . . . . . . . . . . . . . . . . . . . . . . . 22
5.2 Levenes test on equality of variances . . . . . . . . . . . . . . . . . . . . 23
7.1 Premium included Delta . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
7.2 Conversion of a Premium Included Delta to Strike . . . . . . . . . . . . . 34
9.1 Quarterly mean and standard deviation of the goodness of t of Heston
parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
9.2 Quarterly mean and standard deviation of the goodness of t of the
Black-Scholes parameter . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
9.3 Heston parameter values on 1/4/2010 and 6/1/2010 on EURUSD . . . . 44
9.4 Heston parameter values on 1/4/2010 and 6/1/2010 on USDJPY . . . . 44
9.5 Black-Scholes parameter values on 1/4/2010 and 6/1/2010 on EURUSD
and USDJPY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
10.1 Number of options under investigation . . . . . . . . . . . . . . . . . . . 50
10.2 Number of option expirations in quarterly periods . . . . . . . . . . . . . 50
10.3 Delta level on average of shorted EURUSD call options at initiation . . . 51
10.4 Delta level on average of shorted USDJPY call options at initiation . . . 51
10.5 Number of EURUSD call options expiring in-the-money . . . . . . . . . . 53
10.6 Number of USDJPY call options expiring in-the-money . . . . . . . . . . 53
10.7 The mean prot and loss and standard deviation on the hedging error
with Black-Scholes and Heston pricing . . . . . . . . . . . . . . . . . . . 54
v
1
Introduction
In a nancial world which have experienced market crashes starting with Black Mon-
day in 1987 the introduction of extreme market movements have given rise to the
reconsideration of the assumptions behind the pricing of nancial instruments such
as options on stocks as well as foreign exchange. In the past market participants
and practitioners have relied more on the Black- Scholes model and its assump-
tion about asset returns whereas today the market prices of options do not reect
those predicted by the Black-Scholes model. Instead a family of stochastic volatility
models has emerged, with the Heston model being the most well known, with more
realistic assumptions about the probability distribution of asset returns today. Still
though, the Black-Scholes model are applied by market participants and practition-
ers in circumvention that avoid its aws. This thesis incorporates the application
of both types of models and tries to uncover pricing misspecications and, in an
empirical study, investigates if one is preferable to the other given a specic pricing
and hedging setting.
In chapter 3, we start by given an introduction to the FX market and FX plain
vanilla options, which are traded over-the-counter (OTC). This fact inuences the
data collected to represent the market prices, which in this case is retrieved from
Bloomberg where an arbitrage free volatility surface is reported from a collection
of option quotes from several contributors representing the worlds largest nancial
institutions. As opposite to exchange traded options that are quoted with a xed
maturity date and with the initiation of new options only on xed dates, from
Bloomberg we are provided with a full set of new options everyday covering the
same range of maturities just with the expiration one day later than the previous
days quoted options.
1
Chapter 4 covers the Black-Scholes (BS) model and its assumptions about log-
normally distributed asset returns. With specic interest in the pricing of FX op-
tions we present the Garman-Kohlhagen formula, which is a simple extension to
the BS model. In this chapter we furthermore introduce the concept of the im-
plied probability density function and risk neutral valuation. Finally we present the
simulation of the BS model.
In Chapter 5 we analyse the distribution of FX log returns considering a sample
of recent years spot FX rates and compare this with the assumption of log-normal
distributed returns in the BS model. The ndings here inspire to consider dier-
ent assumptions on the distribution of log returns, which leads us to introduce a
stochastic volatility model in the next chapter.
Chapter 6 then introduces the process and the closed form solution to the Heston
model. In the calibration of the Heston model we calibrate to this closed form
solution by numerical integration. Furthermore we present the simulation of the
Heston model that is carried out in a mixing solution framework simulated in a
Milstein scheme.
Before the empirical study we present chapter 7, which explains the very FX
specic quoting conventions. More comprehensive than other option markets the
FX option market has a wide range of possible conventions which need to be properly
handled in order to be able to build a volatility surface based on the quotes in the
market. More specically the volatilities are quoted in trading structures that needs
to be converted. Moreover the options are quoted in terms of Delta in the moneyness
dimension. Depending on the Delta convention of the specic FX pair, we need to
use a numerical estimation technique to retrieve the strike level.
Chapter 8 consists of an overview of the data used in the empirical study.
In chapter 9 we then calibrate the BS model and the Heston model to each
day of 371 trading days in the period from 1/4/2010 - 06/22/2011. We present
the objective function and its inherited weighting scheme that is common for both
models. We furthermore analyse the sensitivity of the volatility surface to the change
in Heston parameters by looking at two dierent days. Also a comparison between
the ability of the two models to t the observed market prices is done by calculating
the goodness of t for each model.
In chapter 10 we lay out the hedging strategy consisting of a dynamic BS Delta
hedge with updating implied volatility simulated in the BS model and simulated in
the Heston model. More specically we hedge a number of shorted call options with
dierent maturities and strike levels. We then identify which elements that change
the value of the hedging portfolio. Finally we present the ndings of the study
comparing the BS model as a tool in the interpolation/extrapolation of the updating
2
implied volatility to the Heston model by comparing the hedging performance of the
same BS Delta hedge.
3
2
Problem Statement
In this study we consider the two FX pairs EURUSD and USDJPY. We start by
the following introductory research questions:
I. How are FX returns distributed considering a period of recent years?
II. How does the distribution of FX returns compare to the assumptions about
log-normal distributed asset returns in the Black-Scholes model?
As pointed out by (Reiswich and Wystrup, 2010), the smile construction proce-
dure and the volatility quoting mechanisms are FX specic and dier signicantly
from other markets...Market participants entering the FX OTC derivative market
are confronted with the fact that the volatility smile is usually not directly observable
in the market...Unlike in other markets, the FX smile is given implicitly as a set of
restrictions implied by market instruments. This lead us to the question:
III. How do we handle the FX specic quoting conventions in order to end up with
market prices on plain vanilla option.
In a very recent paper "Applying hedging strategies to estimate model risk and
provision calculation" (Elices, 2011) the authors study the hedging performance of
the BS model and the Vanna-Volga method by assuming that the market volatility
surface is driven by Hestons dynamics calibrated to market for a given time horizon.
The hedging strategy is then built in order to neutralize the uncertain factors in the
Heston model which consist of the spot and the volatility.
In the same way, we rely on a model dependant building of the volatility surface
by calibration of the BS model and the Heston model, respectively, to the observed
market prices.
4
IV. How well does the Black-Scholes and Heston model, respectively, reect a set
of market prices on plain vanilla options over a recent period?
Then we use these calibrations in order to investigate how well a pure Delta
hedging strategy, with the Delta calculated as a BS Delta, is able to replicate the
payo of a plain vanilla FX call option contract. We create a setting where a
set of European plain vanilla FX options with dierent maturities and strikes are
sold every day during a period of 371 trading days. By delta hedging each option
contract individually until its expiry, we obtain the hedging error that we express as
the dierence between the payo of the option contract and the hedging portfolio.
Two experiments are set up where we calculate the BS Delta dynamically with an
updating volatility from the Black-Scholes model and an updating implied volatility
from the Heston model. This leads to the nal research questions:
V. Applying a dynamic BS Delta hedge with updating implied volatility under
the assumption of Black-Scholes underlying dynamics, what is the standard
deviation of the hedging error for each option contract?
VI. Applying a dynamic BS Delta hedge with updating implied volatility under
the assumption of Heston underlying dynamics, what is the standard deviation
of the hedging error for each option contract?
VII. Are the outcome of the hedging correlated with the market return?
2.1 Research Approach
We point out and argue for our choice of research approach in three areas of the
thesis: The inclusion of two dierent FX pairs, the building of the implied volatility
surface and the range of option prices used to build the implied volatility surface.
We choose to include both the EURUSD and the USDJPY in the study be-
cause of mainly one reason. The quoting conventions for the two pairs are
dierent and by including both we show how to handle these dierent quot-
ing conventions. In addition to this reason, the volatility surface of these
two pairs has historically had dierent shapes with the EURUSD exhibit-
ing more of a symmetrical smile and the USDJPY exhibiting a step skew
(Bossens, Rayee, Skantzos, and Deelstra, 2010), (Beneder and Elkenbracht-
Huizing, 2003), (Chalamandaris and Tsekrekos, 2008). Like other studies this
is an attempt to cover a dierent set of market conditions (Bossens, Rayee,
Skantzos, and Deelstra, 2010).
5
We calibrate to raw data where no interpolation or extrapolation has taken
place beforehand. Alternatively we could have used a SVI parametrisation
(Gatheral, 2006) or some other functional form to rst build the surface and
then calibrate to a set of interpolated/extrapolated prices.
We calibrate to only a few number of options counting 5 dierent maturi-
ties and 5 dierent strike levels. This is done because of two reasons. First,
we want to calibrate only to raw data that has not yet been interpolated
in Bloombergs own interpolation scheme, which can be seen in (Bloomberg,
2011). Bloombergs interpolation is based on ATM, 25 Delta and 10 Delta
quotes and if available also 35 25 and 5 Delta (Bloomberg, 2009). This fact
ensures us that we only calibrate to raw data. Second, a lot of eort has gone
into the development of methods that are able to build the full implied volatil-
ity surface with only a few set of option prices (Malz, 1997), (Castagna and
Mercurio, 2006), (Reiswich and Wystrup, 2010). On an OTC option market,
often only a few prices is available and we want to restrict this study to include
only the prices that are most often available.
This thesis uses the same range of option prices from the same source as
in U. Wystrup and D. Reiswichs article "FX Volatility Smile Construction"
(Reiswich and Wystrup, 2010) by using the ATM, 10D RR, 25D RR, 10D
VWB and 25D VWB quotes published on Bloomberg.
2.2 Delimitation
The thesis is limited in areas where additions would bring more accuracy and detail
into the study.
In order to test a pricing model for its misspecications a classical hedging
experiment like the one carried out in (Bakshi, Cao, and Chen, 1997) and
(Elices, 2011) could be done. Here, they test a models ability to replicate
an option payo by taking positions in all assets necessary to neutralize risk
with that number depending on the assumption of the given pricing model.
For the Heston model this implies taking a position in both the underlying
and another option in order to obtain a delta-neutral hedge. In this thesis we
restrict ourselves to only take a position in one asset, the underlying. So this
study cannot be classied under this type of conventional approach.
The interest rate setting in this study is simplied. There has been no building
of an interest rate term structure to use in the simulation of the option pricing
6
models. Nor have we considered option pricing models with stochastic interest
rates like in (Bakshi, Cao, and Chen, 1997). Also we disregard the topic of
default risk in interest rates which is a warm topic today after the current
nancial crises.
No jump models has been considered like a stochastic volatility plus jump in
the underlying (SVJ) model. These types of models are better at reecting the
volatility surface in the short term in comparison with a stochastic volatility
model (Gatheral, 2006). Considering the two FX pairs included in this study,
and the shape of their respective volatility surfaces, a SVJ model might not
even have been able to improve the pricing t in comparison to a stochastic
vol. model.
Researchers point out a necessary adjustment of the volatility quote on steply
skewed markets (Reiswich and Wystrup, 2010), (Bossens, Rayee, Skantzos,
and Deelstra, 2010), (Castagna, 2010). About the quoting conventions on the
foreign exchange option market and the importance of the specic adjustment
of the vega weighted buttery (VWB) quote, the following is said: ...a market
inconsistency that can safely be disregarded in many situations and congura-
tions of prices, but can have a deep impact on the volatility surface building in
others. (Castagna, 2010, p. 116). We have excluded the estimations of such
an adjustment.
Probably the most important limitation of this study is the number of simu-
lations used. This concerns the simulation of the Heston model and the BS
model in the hedging experiment set up. The precision of the pricing in the
Heston model could be improved by increasing the number of simulations,
resulting in an even better hedging performance, assumingly.
7
3
The FX Market
3.1 FX rate
A foreign exchange rate (FX rate) is the price of one currency in terms of another
currency. The two currencies make a currency pair. As an example this could be
the currency pair labelled EURUSD. This is the euro/US dollars exchange rate and
by the end of the trading day on the 1
th
of May 2011 this was quoted at 1.3196.
This is the convention on how to quote this particular currency cross, but it is
equivalent to USDEUR 0.7578, which is just the reciprocal value of the rst FX
rate. The exchange rate EURUSD denotes how many US dollars are worth 1 euro.
The domestic (numeraire) currency is the US dollar and the foreign (base) currency
is the euro. So generally speaking, the exchange rate is the price of the base currency
in terms of the numeraire currency. The last time a US dollar was worth more than
a euro was on the 4
th
of December 2002 on which day the exchange rate was quoted
at 0.9997. Since after the introduction of euro coins and banknotes on the 1
th
of
January 2002 this has been the only year that the US dollar has been worth more
than the euro, reected in an exchange rate less than 1.0000.
3.2 FX forward contract
The forward contract provides a hedge for someone who wants to lock in the ex-
change rate for a future transaction. The buyer of a forward contract is then guar-
anteed a future exchange rate. The forward price is decided as
F
0
= S
0
e
(r
d
r
f
)T
(3.1)
8
The underlying asset in such contracts is a certain number of units of the foreign
currency. The variable S
0
is dened as the spot price in domestic currency of one
unit of the foreign currency and equivalently F
0
is the forward price in domestic
currency of one unit of the foreign currency. Both domestic and foreign interest
rates are the continuously compounded risk-free interest rates per annum.
3.2.1 Interest rate parity
Equation 3.1 is exactly the interest rate parity, which in its continuous compounding
form is often equated as
F(t, T) = S
t
e
r
f
(Tt)
e
r
d
(Tt)
(3.2)
or by its money market conventions for capitalization and discounting, i.e simple
compounding (Castagna, 2010, p. 7)
F(t, T) = S
t
(1 + r
f
)(T t)
(1 + r
d
)(T t)
(3.3)
where r
f
and r
d
are the risk-free interest rates per annum and (T-t) follow the time
convention of 360 trading days in a year.
According to the interest rate parity, the forward exchange rate of a given cur-
rency pair is determined by the respective risk-free interest rates. As an example,
we consider a holder of one unit of foreign currency. There are two ways that this
can be converted into domestic currency at time T. One is by investing it for (T t)
years at r
f
and at the same time selling a forward contract. Then at time T you
would be obligated to sell the proceeds from the investment to collect domestic cur-
rency. The other possibility is to exchange the foreign currency to domestic in the
spot market and then invest these at r
d
for (T-t) years. In the absence of arbitrage
opportunities equation 3.4 should then hold (Hull, 2008, p. 113), which is exactly
equation 3.2 rewritten.
e
r
f
(Tt)
F
0
= S
0
e
r
d
(Tt)
(3.4)
The interest rate parity presented here is also called the covered interest rate parity
as opposite to the uncovered interest rate parity (Oldeld and Messina, 1977). The
former comes from the fact that the trading strategy is risk-free. This is opposite to
the latter where you as a holder of the foreign currency still invest in r
f
, but instead
9
of simultaneously entering into a forward contract, you instead keep your position
in foreign currency uncovered and exposed to the movement in the exchange rate
from t to (T t).
Empirical research shows that for developed countries, the covered interest rate
parity holds fairly well. Prior to the dismantling of capital controls, and in many
emerging markets today (interpreted as political risk associated with the possibil-
ity of governmental authorities placing restrictions on deposits located in dierent
jurisdictions), the covered interest rate parity is unlikely to hold (Chinn, 2007).
From an option pricing point of view the covered interest parity is an underlying
assumption in one of the option pricing models introduced later on here.
3.3 FX options
FX options are traded Over-The-Counter (OTC) as opposite to exchange traded
options. As a trading platform an exchange serves as a link between a buyer and
a seller. The exchange will be providing bid and ask quotes and will be on either
one or the other end of the transaction. The market making is in this case carried
out by the exchange. In the case of FX options there is no exchange involved in
the transaction. A trade will be processed directly between buyer and seller. In
one setting, one might think of a buyer being a corporation that is trading from
a hedging or speculative point of view and the seller being a bank. On the FX
options market one might think of the banks as market makers providing the prices
on options and other FX derivatives.
In order to hedge a foreign exchange exposure FX options are an alternative to FX
forward contracts.
The payo from a long position in a European call option is
max(S
T
K, 0) (3.5)
and the payo from a long position in a European put option is
max(K S
T
, 0) (3.6)
with S
T
being the spot exchange rate at maturity T of the option and K the agreed
upon strike price.
10
Assuming we have the pair EURUSD, two counterparties entering into a plain
vanilla FX option contract can agree on the following, according to the type of
option traded:
Type EUR call USD put: The buyer has the right to enter at expiry into a
spot contract to buy (sell) the notional amount of EUR (USD), at the strike
FX rate level K.
Type EUR put USD call: The buyer has the right to enter at expiry into a
spot contract to sell (buy) the notional amount of EUR (USD), at the strike
FX rate level K.
Considering, as an example, the last type listed above, an American company
due to receive euro at a known time in the future can hedge its risk by buying put
options on euro that mature at that time. This strategy guarantees that the value
of the euros will not be less than the strike price while still allowing the company to
benet from any favorable upward movements in the exchange rate. Similarly, if the
company where to pay euros in the future they could hedge their expose to upward
movements in the exchange rate by buying calls on euros, the rst type listed above.
whereas forward contracts locks in the exchange rate for a future transaction and
guarantees the parties an exchange rate, as described above, an option provides a
type of insurance. It costs nothing to enter into a forward contract, whereas options
require a premium to be paid paid up front in order to be insured.
11
4
The Black-Scholes model
This chapter reviews the most well-known option pricing model, The Black-Scholes
model (Black and Scholes, 1973), because of its inclusion in the empirical study.
Also it remains the building block of present option pricing models, including the
Heston model and the Bates model.
4.1 Geometric Brownian Motion
Black-Scholes assumes the underlying spot price to follow a geometric Brownian
motion generating log-normally distributed returns, the spot price in this case being
the exchange rate on any given FX pair. The process is stochastic by including a
Wiener process that introduces the randomness to the spot price.
dS
t
= S
t
dt + S
t
dW = S
t
(dt + dW) (4.1)
The spot price S
t
depends on S
t
itself, a constant drift, , a constant volatility term,
, and a standard Wiener process, W
t
, where dt is denoting a time dierential. In
order to obtain the explicit solution to this stochastic dierential equation (SDE) we
consider equation 4.2 the process of logS, i.e. the process describing the log-returns.
dlogS
t
= (
1
2

2
)dt + dZ (4.2)
i.e
logS
T
= logS
0
+ (
1
2

2
)T + dZ (4.3)
12
and the explicit solution is then obtained by taking the exponential of logS
S
T
= S
0
e
(
1
2

2
)T+Z
T
(4.4)
4.2 The Black-Scholes equation
With the empirical study of this thesis in mind we have a look at the derivation of
the Black-Scholes (BS) equation which is governing the BS option pricing formula.
This will tell us the principle of delta hedging. Furthermore we take a look at the
necessary adjustments to the Black Scholes equation in order to be able to price FX
options in particular. As a note it is not in the interest of this thesis to go through
the derivation of the solution to the BS equation that will lead to the BS formula.
The Black-Scholes equation can be derived in many alternative ways i.e. using
empirically established nancial theories such as the CAPM and Arbitrage Pricing
theory. The most general derivation assumes an economy with only the underlying
asset and a risk-free money market deposit/risk-free bond which together makes
up the replicating portfolio of the value of the derivative. Meanwhile, the original
derivation uses what is known as the hedging argument, and that is the derivation
that we will outline here (Rouah, 2011).
The derivation follows from imposing the condition that a risk-free portfolio
made up of a position in the underlying asset and the option on that asset must
return the same interest rate as other risk-free assets. As a result of this Black
and Scholes propose that if it is possible to hedge an option position by dynamically
rebalancing a stock position, then the price of a European call option should depend
on the underlying spot price, S
t
(i.e. the FX rate), and the time to maturity on the
option, T.
In order to perform such a hedge Black and Scholes assumes a set of conditions
to hold that they call the ideal market condition:
The FX rate, S
t
, follows the geometric Brownian motion with known constant
drift, , and volatility, .
The option can be exercised only at maturity.
Trading takes place continuously in time.
Money can be borrowed and lend at the same risk-free interest rate.
Short selling is allowed.
13
Short-term risk-free interest rates (r
d
and r
f
) are known and constant.
The underlying asset pays no dividends. (This assumption is relaxed in the
case of FX options.)
We consider a portfolio made up of a quantity of the risky asset (i.e. the FX
pair) and short one option on the FX pair (a put or a call, not yet specied). Let
f(S, t) denote the value of the option and (t) the value of the portfolio.
(t) = S f(S, t) (4.5)
is chosen at every time t so as to make the portfolio riskless. The self-nancing
assumption implies that
d(t) = dS df(S, t) (4.6)
In order to decide the quantity to meet this condition we want to know the dynam-
ics of f(S, t). Here we use Itos Lemma, which is a rule for calculating dierentials
of quantities dependent on stochastic processes.
df(S, t) =
f
t
dt +
f
S
dS +
1
2

2
S
2

2
f
S
2
dt (4.7)
and by plugging in 4.7 into 4.6 we get
d = dS (
f
t
dt +
f
S
dS +
1
2

2
S
2

2
f
S
2
)dt
(
f
S
)dS (
f
t
+
1
2

2
S
2

2
f
S
2
)dt (4.8)
observing that the term dS is the only risky element to the portfolio value, we can
eliminate this by setting
(
f
S
) = 0
which is satised if
=
f
S
(4.9)
Then we have constructed a risk-free portfolio with the dynamics given in the last
part of 4.8 and by a no arbitrage argument the portfolio must yield the risk-free
interest rate, i.e.
14
d = rdt (4.10)
Plugging the risk-free dynamics of the option value in 4.8 and the rst equation 4.5
into 4.10 and rewrittin, we get the BS equation in 4.11.
(
f
t
+
1
2

2
S
2

2
f
S
2
)dt = r(
f
S
S f(S, t))dt

f
t

1
2

2
S
2

2
f
S
2
= r(
f
S
S f(S, t))
f
t
+
1
2

2
S
2

2
f
S
2
+ r
f
S
S rf = 0 (4.11)
The derivation stipulates that in order to hedge the single option, we need to hold
a quantity of the FX pair, which turns out to be the quantity
f
S
. This is the
principle behind delta hedging. Any price of a derivative with the same assumed
process for the underlying as in equation 4.1 has to follow the BS equation.The
equation has many solutions for the derivative price, f, where the particular price
that is obtained depends on the payo function of the given derivative. In the
case of a European call/put the solution is obtained in the BS formula, but for
more complex payo functions accompanied by more exotic options the analytical
solution may be hard to obtain.
4.3 The Garman-Kohlhagen formula
In the same year 1973 as the Black and Scholes paper was published the pricing
model was quickly adjusted to include dividend paying stocks by Merton (1973).
Robert C. Merton further concludes in this paper that the assumption of log-
normally distributed returns and continuous trading is critical to the model. With-
out these, the delta hedge would not give a perfect hedge, thus making the arbi-
trage argument invalid. Many years later after the FX options was rst listed on
the Philadelphia Stock Exchange in 1982 (Exchange, 2004), the pricing model was
adjusted to also be able to price FX plain vanilla options (Garman and Kohlhagen,
1983). Under similar assumptions as in Black-Scholes, that it is possible to operate
a perfect local hedge between a FX option and underlying foreign exchange, Gar-
man and Kohlhagen derive a PDE. One of the insights is that the risk-free interest
rate of foreign currency r
f
has the same impact on the FX option price as the con-
tinuous dividend yield on the stock option. The main contribution is to combine
the Black-Scholes model with the interest rate parity theory, as presented in the
15
beginning of this thesis. More precisely, by assuming the covered interest rate par-
ity to hold and the underlying FX rate to follow a geometric brownian motion, the
logarithmic dierence between the forward, F(t, T), and the spot, S(t), FX rates
can be explained by the spread between the domestic risk-free interest rate, r
d
, and
the foreign risk-free interest rate, r
f
.
The resulting pricing formula for a call option in equation 4.12 is presented in its
forward rate form, where the forward rate is explicitly present in the formula. This
is a Black model (Black, 1976) (adjusted to price FX options), which is a variation
of the original BS model and can be generalized into a class of models known as
log-normal forward models. The adaption of the covered interest rate parity into
the option pricing formula becomes apparent when we compare the calculation of
the forward rate in Equation 4.12 to Equation 3.2.
c = e
r
d
(t,T))
[F(t, T)(d
1
) K(d
2
)] (4.12)
d
1
=
ln(
F(t,T)
K
) +
1
2

d
2
= d
1

F(t, T) = S
t
e
r
f
(t,T)
e
r
d
(t,T)
with the the equivalent spot rate form of the Garman-Kohlhagen formula
c = S
0
e
r
f
(t,T)
(d
1
) Ke
r
d
(t,T)
(d
2
) (4.13)
d
1
=
ln(
S
0
K
) + (r
d
(t, T) r
f
(t, T) +
1
2

2
)

d
2
= d
1

The foreign and domestic interest rates are risk-free and constant over the term of
the options life. All interest rates are expressed as continuously compounded rates.
4.3.1 Implied Probability Density Functions
In order to establish a link between the observed option prices in the market and the
characteristic shapes of the volatility surface we mention the implied risk-neutral
density function (RND).
16
The RND in the Black-Scholes model is assumed to be lognormal with mean
(r
d
r
f
v
2
/2)(T t) and variance v
2
(T t).
The price of an undiscounted call option is given by
C(S
0
, K, T) = E[max{S
T
K, 0}] (4.14)
=
_

K
(s K) (s; T, S
0
)ds (4.15)
where (s; T, S
0
) in (4.15) is the probability density function of S
T
. This is a general
pricing formula independent of the choice of pricing model. Pricing an option in
this framework requires the knowledge of the probability density function, which is
the distribution of the future spot prices.
(Breeden and Litzenberger, 1978) found that provided a continuum of European
call options with same maturity and a strike range going from zero to innity writ-
ten on a single underlying FX pair, we can recover the RND in a unique way by
dierentiating (4.15) with respect to K twice
C
K
=
_

K
(s; T, S
0
)ds (4.16)

2
C
K
2
= (s; T, S
0
)ds (4.17)
4.3.2 Risk-neutral valuation
Another approach to nd the price of a derivative is by risk neutral valuation or
equivalently by the Martingale approach. The equivalence between the PDE ap-
proach and the risk neutral valuation is guaranteed by Feynman-Kac by establishing
a link between PDEs and stochastic processes.
The solution to the Garman-Kohlhagen equation can also be expressed in terms
of an expectation. By the Feynman-Kac theorem we have
V (S
t
, t) = E
Q
_
e

T
t
r
d
s
ds
V (S
T
, T)
_
(4.18)
where S
t
is the solution to the SDE (4.1) with = r
d
r
f
. The drift is risk
neutral and consists of the continuously compounded domestic interest rate net of
the foreign interest rate. What (4.18) says is that the value of a contingent claim
(a claim that is dependant on the underlying value) like a European option, can be
calculated by nding the risk neutral expectation of the discounted terminal payo.
The terminal payo is discounted by the domestic interest rate and the risk neutral
17
expectation and the Q measure involves the process of S
T
to evolve not as original
but risk neutrally.
To recapitulate the general pricing framework above, there is a connection be-
tween the existence of a replication portfolio replicating the nal value of the op-
tion, and the existence of a equivalent martingale measure. They both guarantee an
arbitrage-free price. This can be calculated as the current value of the replication
portfolio, or as the expected value of the discounted terminal payo of the option
calculated under the risk-neutral probability measure.
4.4 Simulation of the Black-Scholes model
We consider the risk neutral process in Equation (4.19) and compute the risk neutral
expectation of the terminal payo as suggested by the Feyman-Kac theorem.
dS
t
= (r
d
t
r
f
t
)S
t
dt +
t
S
t
dW (4.19)
18
5
Empirical facts
5.1 The distribution of FX returns
Empirically we observe a departure from the normality assumption in the Black-
Scholes model when we have a look at the distribution of log returns on EURUSD
and USDJPY. In gures 5.1 and 5.2 the frequency distributions of two samples of
daily log returns from 1/6/2006 - 5/3/2011 is pictured. A lognormal distribution
with the same mean and standard deviation as the implied distribution is depicted
by the solid line. The empirical distributions are highly peaked compared to the
normal distribution. Furthermore from gures 5.3 and 5.4, which depict a Q-Q
plot of the log returns vs. a normal distribution, we can observe that the empirical
distributions of log returns does in fact exhibit fat tails and clearly deviates from the
normality assumption. From the visual evidence of a highly peaked and fat tailed
distribution (leptokurtic), we can conclude that small and large movements in the
empirical samples occur more likely compared to normally distributed log returns.
By looking at gures 5.5 and 5.6, where we plot the daily log returns of EURUSD
and USDJPY, we see that large moves follow large moves (both up and down) and
small moves follow small moves (both up and down). This is the so-called volatility
clustering, where we observe that high and low volatility is clustered around certain
time periods. This observation indicates autocorrelation, which is conrmed in
Figures ?? - ??. Here the autocorrelations of absolute returns are estimated where
all lags included is signicantly positive. In addition to this, Figures 5.9 and 5.10
demonstrates mean reversion in the log returns by showing how volatility evens out
when measured over a longer horizon.
19
!
#!
$!
%!
&!
'!!
'#!
(
!
)
!
$

(
!
)
!
*
+

(
!
)
!
*
$

(
!
)
!
*
'

(
!
)
!
#
&

(
!
)
!
#
,

(
!
)
!
#
#

(
!
)
!
'
-

(
!
)
!
'
%

(
!
)
!
'
*

(
!
)
!
'

(
!
)
!
!
+

(
!
)
!
!
$

(
!
)
!
!
'

!
)
!
!
#

!
)
!
!
,

!
)
!
!
&

!
)
!
'
'

!
)
!
'
$

!
)
!
'
+

!
)
!
#

!
)
!
#
*

!
)
!
#
%

!
)
!
#
-

!
)
!
*
#

!
)
!
*
,

!
)
!
*
&

!
"
#
$
%
&

(
)
&
*
+
&
,
-
.

/"0%. %123)&4+),
5676!/
!
#!
$!
%!
&!
'!!
'#!
(
!
)
!
$

(
!
)
!
*
%

(
!
)
!
*
#

(
!
)
!
#
&

(
!
)
!
#
$

(
!
)
!
#

(
!
)
!
'
%

(
!
)
!
'
#

(
!
)
!
!
&

(
!
)
!
!
$

!

!
)
!
!
$

!
)
!
!
&

!
)
!
'
#

!
)
!
'
%

!
)
!
#

!
)
!
#
$

!
)
!
#
&

!
)
!
*
#

!
)
!
*
%

!
)
!
$

!
"
#
$
%
&

(
)
&
*
+
&
,
-
.


/"0%. %123)&4+),
5!/678
Figure 5.1: Empirical sample frequency
for EURUSD
Figure 5.2: Empirical sample frequency
for USDJPY
-4 -3 -2 -1 0 1 2 3 4
-0.03
-0.02
-0.01
0
0.01
0.02
0.03
0.04
Standard Normal Quantiles
Q
u
a
n
t
ile
s

o
f

I
n
p
u
t

S
a
m
p
le
QQ Plot of Sample Data versus Standard Normal
-4 -3 -2 -1 0 1 2 3 4
-0.04
-0.03
-0.02
-0.01
0
0.01
0.02
0.03
0.04
0.05
0.06
Standard Normal Quantiles
Q
u
a
n
t
ile
s

o
f

I
n
p
u
t

S
a
m
p
le
QQ Plot of Sample Data versus Standard Normal
Figure 5.3: Q-Q plot for EURUSD Figure 5.4: Q-Q plot for USDJPY
!"#"$
!"#"&
!"#"'
!"#"(
"
"#"(
"#"'
"#"&
"#"$
"#")
"#"*
'""* '""+ '"", '""- '"(" '"((
!
"
#
$
%

$
'
(

)
*
+
,
)
-

.*")
/0102!
!"#"$
!"#"&
!"#"'
!"#"(
"
"#"(
"#"'
"#"&
"#"$
"#")
"#"*
'""* '""+ '"", '""- '"(" '"((
!
"
#
$
%

$
'
(

)
*
+
,
)
-

.*")
/0!12.
Figure 5.5: Daily log returns for EU-
RUSD
Figure 5.6: Daily log returns for USD-
JPY
0 2 4 6 8 10 12 14 16 18 20
-0.2
0
0.2
0.4
0.6
0.8
Lag
S
a
m
p
le

A
u
t
o
c
o
r
r
e
la
t
io
n
Sample Autocorrelation Function
0 2 4 6 8 10 12 14 16 18 20
-0.2
0
0.2
0.4
0.6
0.8
Lag
S
a
m
p
le

A
u
t
o
c
o
r
r
e
la
t
io
n
Sample Autocorrelation Function
Figure 5.7: Autocorrelation for EU-
RUSD
Figure 5.8: Autocorrelation for USD-
JPY
20
!
!#!$
!#%
!#%$
!#&
!#&$
!#'
&!!( &!!) &!!* &!%! &!%%
!
"
#
$
%
&
"
'

)
%
*
+
,
*
"
$
-

./+&
012134
' +,-./ % 0123
!
!#!$
!#%
!#%$
!#&
!#&$
!#'
!#'$
&!!( &!!) &!!* &!%! &!%%
!
"
#
$
%
&
"
'

)
%
*
+
,
*
"
$
-

./+&
01234.
' +,-./ % 0123
Figure 5.9: Rolling historic volatility
for EURUSD
Figure 5.10: Rolling historic volatility
for USDJPY
5.1.1 Jarque-Bera
To conrm our results and to nd further evidence against the normality assumption
underlying the Black-Scholes model we make use of the Jarque-Bera test (Jarque and
Bera, 1987). Based on the sample kurtosis and skewness we test the null hypothesis
that the data is drawn from a normal distribution. The null hypothesis is a joint
hypothesis of the skewness being 0 and the excess kurtosis being 0, which in the
latter case is the same as a kurtosis of 3.
The overall conclusion by looking into tabel 5.1, when considering the full sam-
ple of log returns, is that we clearly reject the null hypothesis, that the sample data
is from a normal distribution, in both the EURUSD and USDJPY case. This con-
clusion comes with a high degree of certainty with a signicance level below 0.1%.
When we then have a look at the separate years considering rst the EURUSD,
we are able to reject in 3 out of 6 years at a signicance level of 5.0%, whereas
for the USDJPY case this is 4 out of 6 years. When looking into the estimates of
the overall skewness and kurtosis and comparing the two pairs, one observes that
in terms of skewness the EURUSD deviates the most from a normal, whereas in
terms of kurtosis it is the USDJPY that deviates the most from the normal. These
dierences in skewness and kurtosis between the two pairs is somewhat visual in
gures 5.1 and 5.2 from before. Comparing the tails of the frequency distributions
one might see that the EURUSD log returns has a longer right tail exhibiting more
positive skewness whereas the USDJPY log returns has a longer left tail exhibiting
more negative skewness (Even though apparently not enough for the full sample to
be negatively skewed). Both distributions though are on an overall scale slightly
positively distributed meaning that most values are concentrated on the left of the
mean, with extreme values to the right (as opposite to negatively skewed distribu-
tions, where most values are concentrated on the right of the mean, with extreme
values to the left). The dierence in the kurtosis of the two pairs of log returns is
also somewhat visual from the gures 5.3 and 5.4 from before, where the USDJPY
21
Table 5.1: Jarque-Bera test on normality
EURUSD USDJPY
period skewness excess kurtosis JB sign. level skewness excess JB sign. level
2006 0.145 0.001 > 50.000% 0.242 0.164 20.473%
2007 0.331 3.573 2.186% 0.712 4.675 < 0.100%
2008 0.330 1.500 < 0.100% 0.269 3.807 < 0.100%
2009 0.193 1.091 0.540% 0.154 0.392 21.547%
2010 0.014 0.237 > 50.000% 0.073 5.270 < 0.100%
2011 0.120 0.007 > 50.000% 0.410 0.941 4.660%
2006-
2011
0.124 1.937 < 0.100% 0.044 4.378 < 0.100%
log returns seems to exhibit the most kurtosis.
The test statistic JB is dened as
JB =
n
6
(S
2
+
1
4
K
2
) (5.1)
where n is the number of observations, S is the sample skewness in Equation 5.2
and K is the sample excess kurtosis in Equation 5.3.
S =

3

3
=
1
n

n
i=1
(x
i
x)
3
(
1
n

n
i=1
(x
i
x)
2
)
3
2
(5.2)
K =

4

4
3 =
1
n

n
i=1
(x
i
x)
4
(
1
n

n
i=1
(x
i
x)
2
)
2
3 (5.3)
where
3
and
4
are the estimates of the third and fourth central moments, respec-
tively, x is the sample mean and is the estimate of the second central moment,
the variance.
22
5.1.2 Levene
Excess kurtosis might indicate heteroscedastic returns, where homoscedastic returns
is the assumption underlying the Black & Scholes model. We therefore perform the
Levenes test of homoscedatic returns, where the null hypothesis is that the variance
of two successive subsamples are equal as well as the variances of all subsamples.
Considering the latter we strongly reject the hypothesis that the variance in the
subsamples are constant thus violating the assumption in the Black Scholes model.
Comparing the individual successive yearly subsamples, in the case of the EURUSD
we are able to reject in 2 out of 5 cases at a signicance level of 5%. In the case
of the USDJPY this is 4 out of 5 cases in correspondence with the superior excess
kurtosis compared to the EURUSD case.
Table 5.2: Levenes test on equality of variances
EURUSD USDJPY
period 1 period 2 volatility 1 volatility 2 Levene sig. level volatility 1 volatility 2 Levene sig. level
2006 2007 7.35% 6.16% 0.859% 7.83% 9.62% 1.244%
2007 2008 6.16% 13.78% 0.000% 9.62% 16.18% 0.000%
2008 2009 13.78% 12.03% 9.691% 16.18% 12.68% 1.659%
2009 2010 12.03% 11.76% 75.421% 12.68% 10.36% 2.458%
2010 2011 11.76% 9.85% 7.890% 10.36% 9.87% 74.257%
2006 2011 0.000% 0.000%
23
6
The Heston model
The most well-known and popular of all stochastic volatility models is the Heston
model (Gatheral, 2006) and was presented in (Heston, 1993).
6.1 The process
The process followed by the underlying asset in the Heston model is
dS
t
= S
t
dt +

v
t
S
t
dW
(1)
t
(6.1)
dv
t
= (v
t
)dt +

v
t
dW
(2)
t
(6.2)
with
dW
(1)
t
dW
(2)
t
= dt
where is the rate of reversion of v
t
to the long run variance, , is the volatility
of volatility and is the correlation between the two stochastic increments of the
processes dW
(1)
t
and dW
(2)
t
. The process of the underlying in (6.1) is the same
process assumed in the Black Scholes model presented in (4.1) only now the volatility
is stochastic. That is, another random factor is introduced by dW
(2)
t
. What denes
the specic process of the underlying in the Heston model compared to the general
case of stochastic volatility models is
dv
t
= (S
t
, v
t
, t)dt + (S
t
, v
t
, t)

v
t
dW
(2)
t
(6.3)
(S
t
, v
t
, t) = (v
t
)
(S
t
, v
t
, t) = 1
24
where the process followed by the instantaneous variance, v
t
, can be categorized as
a version of the square root process (CIR) in (Cox, Ingersoll Jr, and Ross, 1985).
Given that the Feller condition in equation (6.4) is satised the variance process is
always strictly positive. (Anderson, 2005) shows that this condition is often violated
when calibrating the Heston model to market data.
2
2
(6.4)
What makes the Heston stochastic volatility model stand out from other stochastic
volatility models can be adressed to two reasons. First, the volatility process is
non-negative and mean reverting which is what we observe in the market. Secondly,
The Heston model has a semi-analytical closed form solution for European option,
which is fast and relatively easy to implement. The closed form solution is especially
useful when calibrating the parameters in the model to the observed vanilla option
market. This ecient computational ability of the model is characterised as the
greatest advantage of the model over other potentially more realistic SV models
(Janek, Kluge, Weron, and Wystup, 2010).
Furthermore, after adapting the model to a FX setting, the model is described
as being particular useful in explaining the volatility smile found in FX markets
often characterised by a more symmetrical smile when comparing to equity markets
where the structure is a strongly asymmetric skew as a consequence of the leverage
eect on these markets(Janek, Kluge, Weron, and Wystup, 2010).
6.2 The solution
The PDE of the Heston model can be derived using the same approach as when we
derive the PDE for the BS model where standard arbitrage arguments is used. In
addition to the replication portfolio used to derive the BS model another asset in
the form of an option is added in order to hedge the randomness introduced by the
stochastic volatility. The following PDE can then be derived
V
t
+
1
2
vS
2

2
V
S
2
+ vS

2
V
vS
+
1
2

2
v

2
V
v
2
+ rS
V
S
rV
+{( V ) (S, v, t)

v}
V
v
= 0 (6.5)
where (S, v, t) is the market price of volatility risk. The closed-form solution of a
European call option on an FX pair for the Heston model is
S
t
P
1
Ke
(r
d
r
f
)(Tt)
P
2
(6.6)
25
where
P
j
=
1
2
+
1

_

0
Re
_
e
iln(K)
f
j
(x, v
t
, , )
i
_
d (6.7)
f
j
= e
C(,)+D(,)v
t
+ix
C(, ) = (r
d
r
f
)i +
a

2
_
(b
j
i + d) 2ln
_
1 ge
d
1 g
__
D(, ) =
b
j
i + d

2
_
1 e
d
1 ge
d
_
g =
b
j
i + d
b
j
i d
d =
_
(i b
j
)
2

2
(2u
j
i
2
)
u
1
=
1
2
, u
2
=
1
2
, a = , b
1
= , b
2
= +
for j = {1, 2} and x = lnS
0
and where put options can be solved by the put-call
parity in equation (6.8).
C(K, T) = P(K, T) + S
t
e
(r
f
(Tt)
e
(r
d
(Tt)
K (6.8)
The integral part of (6.7) with the integration of f
j
is the reason why it is only a semi
analytical closed-form solution, because the integral only can be evaluated approx-
imately. This can be done with reasonable accuracy using a numerical integration
technique. This includes the calculation of the complex logarithm in C(, ) in
equation (6.7), which can cause numerical instabilities. The problem can be solved
almost entirely if d is replaced by d

= d (Albrecher, Mayer, Schoutens, and Tis-


taert, 2005). Meanwhile this correction of the original pricing formula is likely to
have the most impact of long maturities above 3-5 years, where the problem for
the short and middle term options might not even be detectable (Albrecher, Mayer,
Schoutens, and Tistaert, 2005), (Janek, Kluge, Weron, and Wystup, 2010). We later
calibrate the Heston model to the closed form solution presented in Equation (6.6).
The code of the pricing formula is an exact replicate from (Moodley, 2005).
26
6.3 Simulation of the Heston model
The simulation of the Heston model can be done with the mixing solution approach
to stochastic volatility models by (). The mixing theorem is a way of expressing
the option price in a more complicated model like a stochastic volatility model as
a weighted sum of the option prices in a simpler base model, in this case the BS
model. This can also be expressed as in
C(S
t
, v
t
, ) = E[C
BS
(S
eff

,
_
v
eff

, ] (6.9)
where the option price in the Heston model is expressed as an expectation of the BS
value with a so-called eective spot, S
eff

and eective volatility


_
v
eff

. In order
to get to these eective values we start by writing the explicit solution to equation
(6.1) at maturity, T, in a risk neutral setting and with an FX pair as the underlying
asset:
S
T
= S
0
e
(r
d
r
f
)T
1
2

T
0
v
s
ds+

T
0

v
s
dB
s
(6.10)
where
B
t
= W
t
+
_
1
2
Z
t
, dW
t
dZ
t
= 0 (6.11)
and plugging (6.11) into (6.10) and re-arranging
S
T
= S
eff
T
exp
_
(r
d
r
f
)T
1
2
_
T
0
(1
2
)v
s
ds +
_
T
0
_
(1
2
)v
s
_
dZ
s
(6.12)
where the eective spot is
S
eff
T
= S
0
exp
_

1
2
_
T
0

2
v
s
ds +
_
T
0

v
s
dW
s
_
(6.13)
and we express the eective variance as
v
eff
T
=
1
T
_
T
0
(1
2
)v
s
ds (6.14)
27
with V
t
evolving as the stochastic volatility process specic for the Heston model as
described in equation (6.2).
The benet of this approach is to improve Monte Carlo techniques, where the
mixing theorems increase the eciency of Monte Carlo evaluation of option prices
in stochastic volatility models.
6.3.1 The Milstein scheme
In equations (6.13) - (6.14) two integrals over the process v
t
have to be estimated.
In this relation we use the Milstein scheme in the discretization of the process of v
t
,
which in general terms (Glasserman, 2003) can be written as

X(i +1)

X(i) +a(

X(i))h+b(

X(i))

hZ
i+1
+
1
2
b(

X(i))b

(

X(i))h(Z
2
i+1
1) (6.15)
which in the specic case of The Heston model translates to
v
t+1
= v
i
(v
i
)t +

v
t

tZ
t+1
+

2
4
t(Z
2
t+1
1) (6.16)
where if v
t
= 0 and 4/
2
> 1 then v
t+1
> 0, which means that the occurrence of
a negative variance should be substantially reduced compared to the Euler scheme
(Gatheral, 2006, p. 22).
In addition to this we introduce antithetic variables that runs the same calcula-
tions but with an opposite sign of the random value to speed up the converge of the
simulated price to the true model price. Furthermore we introduce the absorbing
assumption that if v < 0 then v = 0 (Gatheral, 2006, p. 22).
28
7
Market data
As noted in the chapter 2, the volatility quoting mechanisms are FX specic and
the FX smile is given implicitly as a set of restrictions implied by market structures.
7.1 Quoting conventions
It is possible to identify some structures that are very popular amongst professional
market participants:
7.1.1 ATM straddle
The quotes for this structure on standard maturities are the most liquid ones
(Castagna, 2010, p. 16).
I. ATM straddle (ATM STDL): The sum of a (base currency) call and a (base
currency) put struck at the at-the-money level.
This structure here is traded in dierent ways depending on which ATM convention
is used. The rst kind is
ATM spot: The strike is set equal to the FX spot rate. (The strike is inde-
pendent of the maturity on the options.)
ATM forward: The strike is set equal to the FX forward rate. (In this case
the strike is dependent on the maturity of the options, c.f. 3.1.)
ATM Delta neutral STDL: The strike is chosen so that, given the expiry, a
put and a call have the same Delta but with dierent signs. (In this case the
29
strike is also dependent on the maturity of the options because of its inuence
on Delta on the two options.)
The ATM STDL implied volatility quoted in the FX option market uses the last
listed ATM convention; the ATM Delta neutral STDL (ATM DNS). So the implied
vol. in this ATM structure is dependent on the strike level implied by the Deltas
on the two options and should not be confused by the simpler implied vol. based
on just setting the strike equal to the spot rate i.e. the ATM spot convention.
ATM STDL = ATM DNS (7.1)
7.1.2 Risk Reversal
Besides the ATM STDL, there are at least two other structures frequently traded.
One of them is the risk reversal.
2. Risk reversal (RR): One buys a (base currency) call and sells a (base currency)
put both featured with the same absolute Delta (symmetric).
Delta can be chosen equal to any level, but the 25% is the most liquid one (Castagna,
2010, p. 17). So the call and the put entering into the RR will have a strike level
yielding a 25% Delta without considering the sign, which for puts will be negative.
The RR is quoted as the dierence between the implied volatilities of the two op-
tion prices and we indicate this price in volatility as rr. Equation 7.2 denotes this
relationship.
RR = Call Put
rr(t, T; 25) =
25C
(t.T)
25P
(t, T) (7.2)
where (t, T) is the implied vol. at t for an option expiring in T and with a strike
corresponding to the Delta level indicated in the subscript.
7.1.3 Vega-weighted buttery
The other structure that is traded frequently is the Vega-weighted buttery.
3. Vega weighted buttery (VWB): One sells an ATM STDL and buys a strangle.
The strangle is the sum of a (base currency) call and put both featured with
the same absolute Delta (symmetric).
30
The 25% Delta VWB is the most traded VWB (Castagna, 2010, p. 18). The VWB
is quoted as the implied volatilities of the two options in the strangle, which are
evenly split, and the implied vol. of an ATM DNS.
V WB = 0.5(Call + Put) ATM DNS
vwb(t, T; 25) = 0.5(
25C
(t, T) +
25P
(t, T))
ATM DNS
(t, T) (7.3)
In the interbank market the quotations of the VWB appear as in 7.3 (Bloomberg).
7.2 Retrieving the implied volatility
From the three main structures dealt on the FX options market, ATM STDL, RR
and the VWB it is possible, from the relationships in equations 7.1, 7.2 and 7.3
to immediately retrieve the implied volatilities. As for the at-the-money implied
vol. it is synonymous with the quoted ATM STDL (from now on instead of writing

ATM DNS
we write simply
ATM
), whereas the implied vol. on a 25 Delta call and
put (actually -25 Delta for the put) can be calculated as in equations 7.5 and 7.6,
respectively.

ATM
= ATM STDL (7.4)

25C
(t, T) =
ATM
(t, T) + vwb(t, T; 25) + 0.5rr(t, T; 25) (7.5)

25P
(t, T) =
ATM
(t, T) + vwb(t, T; 25) 0.5rr(t, T; 25) (7.6)
These examples show the calculations of the implied volatility on a 25D call and
put, but applies to any Delta level assuming the corresponding VWB and RR are
available.
7.2.1 The Delta-sticky convention
FX derivatives markets quote the strike prices in terms of Delta of the option. This
is referred to as the so-called Delta-sticky convention. Its implications concerns the
following:
The sticky Delta: Once the deal is closed, given the level of the FX spot rate
and the implied vol. agreed upon (where the interest rate levels will be taken
from the money market), the strike will be set at a level yielding the BS Delta
that the two counterparties were dealing.
31
Practically this means that, if the FX spot rate moves, all other things being equal,
the curve of implied vol. vs. Delta will remain unchanged, while the curve of the
implied vol. vs. strike will shift.
When the option is quoted with reference to a strike level expressed in terms of
Delta, once the option is traded and the FX spot reference rate and the implied vol.
are xed, then the absolute level of the strike can be retrieved by setting
wP
f
(t, T)(wd
1
) =

(7.7)
which can be expressed as
K = F(t, T) exp
_
w
_
(T t)
1
(|

|/P
f
(t, T)) + 0.5
2
(T t)
_
(7.8)
where the at-the-money strike reduces to
K
ATM
(t, T) = F(t, T) exp
_
0.5
2
ATM
(T t)
_
(7.9)
where w = 1 (respectively, w = -1) if a call (put) option.
1
is the inverse of the
cumulative normal distribution function, and the values and

are the required
inputs. So if one wants to nd the corresponding strike level to a 25 Delta call, the
inputs into equation 7.8 must be the at the 25 Delta call level and

would be
0.25. The latter enters into the formula as its absolute value, which is relevant when
considering put options.
7.2.1.1 Premium included Delta
The Delta convention used in the market for EURUSD and USDJPY has impli-
cations on which method to use when converting Deltas into strikes. In the case
of the EURUSD a regular Delta is quoted whereas in the case of the USDJPY a
premium included/adjusted Delta is the market convention on how to quote Delta
(Bloomberg), (Reiswich and Wystrup, 2010).
In order to explain the dierence between a regular premium excluded Delta
and a premium included Delta we use numbers from the example in (Reiswich and
Wystrup, 2010, p. 4) to create Tabel 7.1.
The relationship between the premium included Delta and Delta is
PI =
V
S
0
(7.10)
32
Table 7.1: Premium included Delta
EURUSD Call
Notional 1,000,000 EUR
S
0
1.3900
K 1.3500
-
T -
Premium 102,400 USD 73,669 EUR
60% 600,000 EUR
PI 52.63% 526,331 EUR
and T is not provided, but is also irrelevant in the example.
where the amount of foreign currency units to buy in a hedge of a short position is
FOR = N(
V
S
0
) (7.11)
If we go short in a EURUSD call option with notional EUR 1 mio. we receive a
premium of USD 102,400, which corresponds to EUR 73,669. Lets say the Delta
on the option is 60%. Then we have to buy EUR 0.6 mio. in order to keep a
local Delta hedge. But considering that we receive something from the trade of the
option, the EUR amount to hedge is only EUR 526,331, which corresponds to a
premium included Delta of 52.63%.
As just mentioned in the EURUSD case, the market convention is actually to
quote the regular premium excluded Delta. So in this case we can rely on equations
7.8 - 7.9 and directly retrieve the strike. This is not the case for the USDJPY where
we have to resort to a numerical procedure (based on the Newton-Raphson scheme),
when we want to retrieve the strike, since the option premium entering into Delta
is a function of the strike itself (Castagna, 2010, p. 35-36). The procedure used in
this study follows this scheme and is outlined in the Appendix.
An example of the conversion of the premium included delta to strike is presented
in Tabel 9.1 where we calculate both the strike retrieved directly from equations
7.8 - 7.9 and the strike adjusted for the premium included (PI) Delta, calculated
numerically, which in this case for the USDJPY is the correct way to retrieve the
strike. This is done both for a call and a put.
The premium included Delta will always be less than the regular premium ex-
cluded Delta. This is true both when we consider calls and puts (where the Delta on
a call is measured along the scale from 0% - 100% and the Delta on a put is measured
along the scale from 0% - (-100%) going from OTM - ITM on both scales). This fact
would imply that, in the case of a call, the strike retrieved from a PI Delta, without
making any adjustments (calculated from 7.8 - 7.9), will always be higher than the
33
Table 7.2: Conversion of a Premium Included Delta to Strike
USDJPY Call USDJPY Put
S
0
81.54 S
0
81.54

25c
11.97%
25p
13.13%
T-t 0.5 T-t 0.5
P
d
0.99828 P
d
0.99828
P
f
0.99786 P
f
0.99786
K 86.596 K 76.905
K adjusted for PI D 86.302 K adjusted for PI D 76.597
4/28/2011
strike adjusted for the PI Delta, calculated numerically. This implies a strike that
is more ITM. In case of a put, the strike retrieved from a PI Delta, without making
any adjustments, will always be higher than the strike adjusted for the PI Delta,
which implies a strike that is less ITM as opposite to the case of the call. This
relationship is apparent in Tabel 9.1 where K > K adjusted for PI D in both the
call and put cases.
34
8
Data description
I. In Chapter 5, Empirical facts:
Daily spot FX rates from 1/6/2006 - 5/3/2011 on 1.388 weekdays on the
EURUSD and the USDJPY.
2. In Chapter 9, Calibration of the models and Chapter 10, Empirical study on
the hedging performance:
Bid and Ask prices on D10 RR, D10 VWB, D25 RR, D25 VWB and the ATM
DNS with maturites 1M, 2M, 3M, 6M and 1Y from 1/4/2010 - 06/22/2011
consisting of 18.550 quotes on 371 trading days for both the EURUSD and the
USDJPY. The conventions used in the quoting is for the ATM setting: the ATM
DNS, for the Delta Premium: Excluded in the case of the EURUSD and Included in
the case of the USDJPY, for the Delta style: Spot Delta (up to < 1Y then forward
Delta), for the RR: Call Put and for the VWB: (Call + Put)/2 ATM DNS.
Preference: Bloomberg BGN and Cuto: New York 10:00.
Daily spot FX rates from 1/4/2010 - 6/22/2011 on 371 trading days.
Source: Bloomberg
Domestic and Foreign interest rates on 1M, 2M, 3M, 6M and 1Y from 1/4/2010
- 6/22/2011 on 371 trading days. More specically, we use the deposit interest
rates Euribor for the EUR, Libor USD for USD and Libor JPY for JPY.
Source: Nordea Analytics
35
9
Calibration of the models
9.1 Building the market implied volatility surface
We want to go from the data as described in Chapter 8, consisting of 5 structures
on each of 5 maturities, to a set of market call option prices with corresponding
strike prices. We calibrate to prices and not imp. vols. Furthermore we recognise
that we have only the OTM part of the IVS for call options after retrieving the imp.
vol. on a 25D call and 10D call by Equation (7.5). In order to get the ITM part of
the IVS for call options we have to make an assumption. The following steps are
carried out in the procedure to convert our market data into market prices on call
options covering a wide range of the IVS:
1. We apply Equations 7.5 - 7.6 in order to get from D10 RR, D10 VWB, D25
RR, D25 VWB and the ATM DNS to
10C
,
10P
,
25C
,
25P
and
ATM
losing
the time subscript.
2. From here we retrieve the strikes from Equations (7.8) - (7.9) in case of the
EURUSD and use the numerical method applied in Table 9.1 in case of the
USDJPY in order to get from a Delta moneyness to strike prices. We use
Equation (7.9) in both the EURUSD and USDJPY case to retrieve
ATM
as
prescribed in ??.
3. We then assume the put-call parity in 6.8 to hold all though research shows
that this is rarely the case (Chalamandaris and Tsekrekos, 2008). Under this
assumption we calculate the BS call prices on the 5 pairs of
10P
/ K

10P
,
25P
/ K

25P
,
ATM
/ K

ATM
,
25C
/ K

25C
,
10C
/ K

10C
. It is the convention on
every option market that the imp. vol. quoted is the BS imp. vol. which
allows us to use the BS model to calculate the prices.
36
We then end up with market call option prices with the Delta moneyness 90D,
75D, (50)D, 25D and 10D with corresponding strike prices.
9.2 Calibration of the Heston model
"The price to pay for more realistic models is the increased complexity of model
calibration: as noted by [Jacquier & Jarrow (2000)], in presence of complex models
the estimation method becomes as crucial as the model itself" (Hamida and Cont,
2005, p. 3) .
The calibration of the parameters in a SV model can be done in two conceptually
dierent ways.
First, one can choose to look at the historical time series of asset returns recording
only the historical spot prices and on this basis try to estimate the model parameters
that would yield the current observed option prices. Applied estimation methods
here have been Generalized, Simulated and Ecient Methods of Moments (Janek,
Kluge, Weron, and Wystup, 2010). In this case 6 parameters incl. has to be
estimated. It has been showed that those parameters that produce the current
observed prices and their time-series estimate counterparties are in fact dierent
(Bakshi, Cao, and Chen, 1997), which makes this way of calibrating a less attractive
choice.
The second way is to calibrate the model parameters to current observed vanilla
option prices. In this case we only have to estimate 5 parameters with the ex-
clusion of . This way of calibrating is also categorized as an indirect method of
approximating the RND by tting the parameters, driving the stochastic process of
the underlying, to observed option prices (Jackwerth, 1999), (Brunner and Hafner,
2003).
We calibrate to the semi analytical closed-form solution in equation (6.7) where
the price of a call option is calculated by numerical integration by approximation of
the integral with the MATLAB function quad, that uses a low order method using
an adaptive recursive Simpsons rule. Alternatively the pricing formula could be
solved by the Fast Fourier Transform (Carr and Madan, 1999).
9.2.1 The calibration problem
The minimization of the sum of squared errors, S(), is not a straight forward prob-
lem. Minimizing the objective function in (9.3) is a nonlinear problem subject to the
nonlinear constraint in (6.4) and is far from being convex. It turns out that usually
37
there exist many local minima (Mikhailov and Nogel, 2003, p. 3). The optimization
optimizers in MATLAB are all local minimum optimizers as fminsearch and lsqnon-
lin. So when applying these functions one is not sure that the solution obtained is
a global minimum. Meanwhile we use the function lsqnonlin to minimize the objec-
tive function, while still trying to get around the potential problem of the existence
of many local minima. This function uses the Trust-Region-Reective Algorithm.
Using the function acquires an initial guess on the parameters at which values the
optimizer will start its search for a local minimum. This has the implication that
the solution is dependant on the initial guess.
In order to try to get around the problem we divide each of the 5 parameters in
a number of values which will create a 5-dimensional space of initial guesses. As an
example is divided into the 3 values = [0.9, 0.45, 0] (which is based on the
knowledge that the spot price and the volatility are in fact negatively correlated). In
this way, by given each parameter 3 values that is tested up against each combination
of the 3 possible values of the 4 other parameters, we will be equipped with 3
5
= 243
initial guesses. This procedure is only carried out for the 1. sample day. The
parameters in the solution with the smallest S() is then chosen as the solution on
the 1. day and carried over as the initial guess at the next day. The solution on the
2. day is then carried over as the initial guess at the 3. day in continuation until
the last day. The function lsqnonlin also allows for an upper and lower bound on
parameters in the solution.
9.2.2 Implementing the Feller condition
The Feller condition in (6.4) is reformulated into
2
2
> 0. (9.1)
The LHS of equation (9.1) is then the rst parameter that goes into the function
that need to be optimized by the 5 parameters. Then, from the solution, we can
retrieve the parameter by setting
=
(2
2
) +
2
2
(9.2)
9.3 Calibration of the Black-Scholes Model
In the BS model there is only one parameter to calibrate, the volatility . This
parameter is by the model specication a xed parameter across the moneyness-
maturity level suggesting a at and constant imp. vol. surface. We here use the
38
exact same objective function in Equation (9.3) as when calibrating the Heston
model, including the weighting scheme of bid and ask prices. That is, we minimize
the objective function in order to nd the best t to the 25 option prices observed
on each day to get one volatility parameter per day.
9.4 Objective Function
The optimization algorithm applied in this study is shown in (9.3).

= arg min

_
1
25
5

i=1
5

j=1
[w
i,j
(C
market
(K
i,j
, T
i
) C
model
(K
i,j
, T
i
| ))]
2
_
(9.3)
We minimize the sum of squared errors between observed option prices covering a
surface in the strike and maturity dimension. More specically we choose to calibrate
to call (mid) prices with strikes corresponding to 90D, 75D, ATM, 25D, 10D within
a given maturity, where the specic maturities are 1 month, 2 months, 3 months,
6 months and 1 year. This gives us 25 (mid) prices on each day governed by the
parameter set , , v
0
, and in the Heston model and in the BS model, which we
estimate by the calibration. Furthermore we implement a weighting scheme, which
we will introduce in the next section.
9.4.1 Weighting scheme
We implicitly introduce weights when choosing which strikes and maturities to cali-
brate the model to. In the case of the strike points things to consider in this respect
are the coverage of the strike dimension. By choosing strike points with a range in
the moneyness dimension from D90% to D10% we cover 90 % of the smile, which
would imply to us the curvature and skewness of the smile. Also the strikes D75,
ATM and D25 are the most traded as mentioned earlier and including the D90%
and D10% these are the once quoted by Nordea and most easily retrievable on
Bloomberg. Considering the maturity points, a prior analysis of the liquidity on
these could be done, if this information was retrievable. Meanwhile we stick to the
very commonly quoted maturities that we have chosen.
The weighting scheme implemented in (9.3) considers the bid-ask spread on the
prices observed in the market like in (Moodley, 2005). What we want to do is
to incorporate the extra information that we get from a bid-ask spread into our
estimation. So our objective is to t better the price with a smaller bid-ask spread
39
compared to a price with a bigger bid-ask spread. The way this is implemented is
apparent in (9.4).
w
i,j
=
1
| C
bid
market
(K
i,j
, T
i
) C
ask
market
(K
i,j
, T
i
) |
(9.4)
So a smaller bid-ask spread will introduce a bigger weight making the error cal-
culated bigger and vica versa, a higher bid-ask spread will give a smaller weight
resulting in a smaller error. The implications of this scheme is that we are tting
better the prices with smaller bid-ask spreads by not allowing the t to get too far
from the mid price, while at the same time allowing the t to move further away
from the mid price with a larger bid-ask spread.
9.5 Calibration results
9.5.1 Goodness of t
In order to get an idea of just how well our calibrations of the models are, we
calculate a goodness of t measure according to equation (9.3). The mean and
standard deviation of the sum of squared errors, S() is calculated on a quarterly
basis as well as overall in Tables 9.1 - 9.2. The numbers are not directly comparable
between the two FX pair calibrations because of the dierence in the level of op-
tion prices as a consequence of very dierent levels of underlying FX exchange rates.
Considering rst the overall Heston calibration to the EURUSD from the Figures
9.6 - 9.7 and Figures 9.10 - 9.11 the calibration performance looks reasonable, all
though not great, when comparing the market prices with the resulting closed form
model prices. Equivalently the Figures from 9.14 - 9.15 and 9.18 - 9.19 provides us
with an idea about the Heston calibration to the USDJPY, which also in this case
seem reasonable but not great. This problem is partly caused by the likeliness that
some or all calibrations have not been able to nd the true global minimum of the
optimization. This problem could be overcome by more advanced methods such as
introducing a heuristic optimization optimizer such as Dierential Evolution as used
in (Gilli, Groe, and Schumann, 2010), also used here in a minimization problem,
allthough a dierent one.
When comparing the goodness of t of the Heston model to that of the BS model
from Tables 9.1 - 9.2, we observe a 10 times better t of the Heston model in the
EURUSD case with almost similar standard deviation around this measure. In the
40
case of the USDJPY the dierence between the t of the Heston model and the BS
model is much more. Here we observe a 50 times better t of the Heston model
with a lot of deviation around the poor t of the BS model suggesting days of even
worse t.
From looking at the volatility smiles through Figures 9.6 - 9.21, we would expect
the BS model to have a better t to the EURUSD than the USDJPY. The former
has a curve closer to a constant at curve which is the only shape that the BS model
can generate.
Table 9.1: Quarterly mean and standard deviation of the goodness of t of Heston
parameters
EURUSD USDJPY
mean std. dev. mean std. dev.
2010 Q1 0.0046 0.00080 0.2577 0.08418
2010 Q2 0.0055 0.00148 0.2278 0.04438
2010 Q3 0.0037 0.00010 0.3524 0.39885
2010 Q4 0.0059 0.00400 0.2094 0.07975
2011 Q1 0.0039 0.00130 0.1293 0.08029
2011 Q2 0.0842 0.07644 0.0963 0.01747
0.0169 0.05188 0.2144 0.19531
2011 Q2 contains the trading days up to 06/22/2011
Table 9.2: Quarterly mean and standard deviation of the goodness of t of the
Black-Scholes parameter
EURUSD USDJPY
mean std. dev. mean std. dev.
2010 Q1 0.1163 0.0264 8.334 2.6798
2010 Q2 0.0996 0.0269 8.984 3.3921
2010 Q3 0.0922 0.0282 9.333 2.5691
2010 Q4 0.0738 0.0404 12.530 3.9771
2011 Q1 0.1428 0.0669 15.206 4.7138
2011 Q2 0.1960 0.0760 15.331 2.8921
0.1186 0.0626 11.564 4.5013
2011 Q2 contains the trading days up to 06/22/2011
Supported by the results in (Bakshi, Cao, and Chen, 1997) we also nd that the
implied parameters in the stochastic volatility model vary through time. This nding
would lead to the conclusion that the model is misspecied. In the case of the BS
model we nd that this model is misspecied because of the presence of a volatility
surface as we see it if we as an example compare Figures 9.8 and Figures 9.9. In
both gures we observe a strike dependence of the imp. vol. and by comparing
the two dierent maturities we further observe a maturity dependence with higher
imp. vol. in the long maturity of 1Y. Furthermore this surface changes dynamically
41
through time, which can be observed by comparing the set of Figures 9.8 and 9.9
to the set of Figures 9.12 and 9.13, where the surface is at a much higher level.
The stochastic volatility model provides us with a more realistic representation of
the underlying dynamics by introducing the possibility of a return distribution that
provides dierent levels of skewness and kurtosis. But the fact that the implied
parameters change through time and therefore has to be re-calibrated is evidence of
its misspecication. These changes in the calibrated parameter values can be seen
in Figures 9.1 - 9.5.
42
!
#
$!
$#
%!
%#
!
&
'
!
$
'
%
!
$
!

!
&
'
!
(
'
%
!
$
!

!
&
'
!
)
'
%
!
$
!

!
&
'
$
!
'
%
!
$
!

!
&
'
!
$
'
%
!
$
$

!
&
'
!
(
'
%
!
$
$

!
"

$
%
&
&
%

'()
*+,+-. +-./01
!
!#!!$
!#!%
!#!%$
!#!&
!#!&$
!#!'
!#!'$
!
(
)
!
%
)
&
!
%
!

!
(
)
!
*
)
&
!
%
!

!
(
)
!
+
)
&
!
%
!

!
(
)
%
!
)
&
!
%
!

!
(
)
!
%
)
&
!
%
%

!
(
)
!
*
)
&
!
%
%

!
"

$
%
&
$
'

()&
,-.-/0 -/0123
Figure 9.1: One week moving average
of
Figure 9.2: One week moving average
of
!"!!
!"$!
!"%!
!"&!
!"'!
("!!
("$!
!
'
)
!
(
)
$
!
(
!

!
'
)
!
%
)
$
!
(
!

!
'
)
!
*
)
$
!
(
!

!
'
)
(
!
)
$
!
(
!

!
'
)
!
(
)
$
!
(
(

!
'
)
!
%
)
$
!
(
(

!
"

$
%
&

'($
+,-,./ ,./012
!"#$
!"#&
!"#'
!"#(
!"#)
!"#*
"
"
+
,
"
*
,
)
"
*
"

"
+
,
"
'
,
)
"
*
"

"
+
,
"
-
,
)
"
*
"

"
+
,
*
"
,
)
"
*
"

"
+
,
"
*
,
)
"
*
*

"
+
,
"
'
,
)
"
*
*

!
"

$
%
&

'()
./0/12 /12345
Figure 9.3: One week moving average
of
Figure 9.4: One week moving average
of
!
!#!!$
!#!%
!#!%$
!#!&
!#!&$
!#!'
!#!'$
!
(
)
!
%
)
&
!
%
!

!
(
)
!
*
)
&
!
%
!

!
(
)
!
+
)
&
!
%
!

!
(
)
%
!
)
&
!
%
!

!
(
)
!
%
)
&
!
%
%

!
(
)
!
*
)
&
!
%
%

!
"
#

%
&
%
'
(
)

!
(
*
%
(
&
+
,

'-,
,-.-/0 -/0123
Figure 9.5: One week moving average of v
t
9.5.2 Sensitivity of skew shape on Heston parameters
The Figures from 9.6 - 9.21 represents two dierent days of option prices with an
interval of half a year and corresponding vol. surfaces for each of EURUSD and US-
DJPY. The maturity dimension is represented by the two options with the shortest
and longest maturity used in the calibration, 1M and 1Y, giving an indication of
the dependence of maturity on the level of imp. volatility. The strike dimension is
made up of all ve strike levels from the calibration.
43
What characterize both surfaces is the level of skewness of the smile in the short
term and long term. Considering rst the imp. vol. on day 1/4/2010 the skew
on both pairs is more pronounced in the long term with higher dierence between
the imp. vol. on deep ITM and far OTM options, with the level of the surface
being higher on the USDJPY. Moving half a year further ahead to the next day the
USDJPY almost maintain this level with an increase in the negative shape of the
skew. A lot more happens in the shape of the EURUSD surface. First, the general
level moves up and the skew becomes almost similar in the short and long term.
These characteristics can to some extend be traced in the dierent parameter
values that we have calibrated on these two days in Tables 9.3 -9.4. But before
comparing the specic parameter values with the Figures 9.6 - 9.21 we take a more
general look at the relationship between the shape of a vol. surface and the Heston
parameters.
Table 9.3: Heston parameter values on 1/4/2010 and 6/1/2010 on EURUSD
EURUSD v
0
1/4/2010 14.31 0.018 0.72 0.31 0.011
6/1/2010 4.42 0.024 0.46 0.45 0.029
Table 9.4: Heston parameter values on 1/4/2010 and 6/1/2010 on USDJPY
USDJPY v
0
1/4/2010 13.41 0.024 0.78 0.22 0.019
6/1/2010 3.93 0.025 0.43 0.49 0.019
Table 9.5: Black-Scholes parameter values on 1/4/2010 and 6/1/2010 on EURUSD
and USDJPY
EURUSD USDJPY

1/4/2010 0.1232 0.1453
6/1/2010 0.1487 0.1348
44
!
!#!$
!#!%
!#!&
!#!'
!#!(
!#!)
!#!*
!#!+
!#!,
!#$
!#!! !#%( !#(! !#*( $#!!
!
"
#
$
%

'%()*
+,-,.'/01
-./012
314256 785419 :5/-
314256 4;-
<= 785419 :5/-
<= 4;-
!"!!
!"!$
!"%!
!"%$
!"&!
!"&$
!"'!
! !"&$ !"$ !"($ %
!
"
#
$
%

'%()*
+,-,.'/01
)*+,-.
/-0.12 3410-5 61+)
/-0.12 07)
89 3410-5 61+)
89 07)
Figure 9.6: Call prices 1M on EURUSD
1/4/2010
Figure 9.7: Call prices 1Y on EURUSD
1/4/2010
!"#
!"##
!"#%
!"#&
!"#'
!"#(
!"#)
!"#*
!"#+
!"#,
!"%
!"%#
!"!! !"%( !"(! !"*( #"!!
!
"
#
$
%
&
'

)
*
$
+
,
$
%
-
.

/&$-+
01213/456
-./012
314256 785419 :5/-
314256 4;-
<= 785419 :5/-
<= 4;-
!"#
!"##
!"#%
!"#&
!"#'
!"#(
!"#)
!"#*
!"#+
!"#,
!"%
!"%#
!"!! !"%( !"(! !"*( #"!!
!
"
#
$
%
&
'

)
*
$
+
,
$
%
-
.

/&$-+
01213/456
-./012
314256 785419 :5/-
314256 4;-
<= 785419 :5/-
<= 4;-
Figure 9.8: Imp. vol. 1M on EURUSD
1/4/2010
Figure 9.9: Imp. vol. 1Y on EURUSD
1/4/2010
!"!!
!"!$
!"!%
!"!&
!"!'
!"!(
!"!)
!"!*
!"!+
!"!,
!"$!
!"!! !"%( !"(! !"*( $"!!
!
"
#
$
%
&
'

)
*
$
+
,
$
%
-
.

/&$-+
01213/456
-./012
314256 785419 :5/-
314256 4;-
<= 785419 :5/-
<= 4;-
!"!!
!"!$
!"%!
!"%$
!"&!
!"&$
!"'!
!"!! !"&$ !"$! !"($ %"!!
!
"
#
$
%
&
'

)
*
$
+
,
$
%
-
.

/&$-+
01213/456
)*+,-.
/-0.12 3410-5 61+)
/-0.12 07)
89 3410-5 61+)
89 07)
Figure 9.10: Call prices 1M on EU-
RUSD 6/1/2010
Figure 9.11: Call prices 1Y on EU-
RUSD 6/1/2010
!"#!
!"##
!"#%
!"#&
!"#'
!"#(
!"#)
!"#*
!"#+
!"#,
!"%!
!"%#
!"!! !"%( !"(! !"*( #"!!
!
"
#
$
%
&
'

)
*
$
+
,
$
%
-
.

/&$-+
01213/456
-./012
314256 785419 :5/-
314256 4;-
<= 785419 :5/-
<= 4;-
!"#!
!"##
!"#%
!"#&
!"#'
!"#(
!"#)
!"#*
!"#+
!"#,
!"%!
!"%#
!"!! !"%( !"(! !"*( #"!!
!
"
#
$
%
&
'

)
*
$
+
,
$
%
-
.

/&$-+
01213/456
-./012
314256 785419 :5/-
314256 4;-
<= 785419 :5/-
<= 4;-
Figure 9.12: Imp. vol. 1M on EURUSD
6/1/2010
Figure 9.13: Imp. vol. 1Y on EURUSD
6/1/2010
45
!"!!
$"!!
%"!!
&"!!
'"!!
("!!
)"!!
!"!! !"%( !"(! !"*( $"!!
!
"
#
$
%
&
'

)
*
$
+
,
$
%
-
.

/&$-+
01/234567
+,-./0
1/2034 5632/7 83-+
1/2034 29+
:; 5632/7 83-+
:; 29+
!"!!
$"!!
%!"!!
%$"!!
&!"!!
&$"!!
!"!! !"&$ !"$! !"'$ %"!!
!
"
#
$
%
&
'

)
*
$
+
,
$
%
-
.

/&$-+
01/234564
()*+,-
.,/-01 230/,4 50*(
.,/-01 /6(
78 230/,4 50*(
78 /6(
Figure 9.14: Call prices 1M on USD-
JPY 1/4/2010
Figure 9.15: Call prices 1Y on USDJPY
1/4/2010
!"#!
!"##
!"#%
!"#&
!"#'
!"#(
!"#)
!"#*
!"#+
!"#,
!"%!
!"%#
!"!! !"%( !"(! !"*( #"!!
!
"
#
$
%
&
'

)
*
$
+
,
$
%
-
.

/&$-+
01/234567
-./012
314256 785419 :5/-
314256 4;-
<= 785419 :5/-
<= 4;-
!"#!
!"##
!"#%
!"#&
!"#'
!"#(
!"#)
!"#*
!"#+
!"#,
!"%!
!"%#
!"!! !"%( !"(! !"*( #"!!
!
"
#
$
%
&
'

)
*
$
+
,
$
%
-
.

/&$-+
01/234564
-./012
314256 785419 :5/-
314256 4;-
<= 785419 :5/-
<= 4;-
Figure 9.16: Imp. vol. 1M on USDJPY
1/4/2010
Figure 9.17: Imp. vol. 1Y on USDJPY
1/4/2010
!"!!
$"!!
%"!!
&"!!
'"!!
("!!
)"!!
!"!! !"%( !"(! !"*( $"!!
!
"
#
$
%
&
'

)
*
$
+
,
$
%
-
.

/&$-+
01/234567
+,-./0
1/2034 5632/7 83-+
1/2034 29+
:; 5632/7 83-+
:; 29+
!"!!
$"!!
%!"!!
%$"!!
&!"!!
&$"!!
!"!! !"&$ !"$! !"'$ %"!!
!
"
#
$
%
&
'

)
*
$
+
,
$
%
-
.

/&$-+
01/234564
()*+,-
.,/-01 230/,4 50*(
.,/-01 /6(
78 230/,4 50*(
78 /6(
Figure 9.18: Call prices 1M on USD-
JPY 6/1/2010
Figure 9.19: Call prices 1Y on USDJPY
6/1/2010
!"#!
!"##
!"#%
!"#&
!"#'
!"#(
!"#)
!"#*
!"#+
!"#,
!"%!
!"%#
!"!! !"%( !"(! !"*( #"!!
!
"
#
$
%
&
'

)
*
$
+
,
$
%
-
.

/&$-+
01/234567
-./012
314256 785419 :5/-
314256 4;-
<= 785419 :5/-
<= 4;-
!"#!
!"##
!"#%
!"#&
!"#'
!"#(
!"#)
!"#*
!"#+
!"#,
!"%!
!"%#
!"!! !"%( !"(! !"*( #"!!
!
"
#
$
%
&
'

)
*
$
+
,
$
%
-
.

/&$-+
01/234564

-./012
314256 785419 :5/-
314256 4;-
<= 785419 :5/-
<= 4;-
Figure 9.20: Imp. vol. 1M on USDJPY
6/1/2010
Figure 9.21: Imp. vol. 1Y on USDJPY
6/1/2010
46
: Changes in changes the degree of symmetry of the smile. Going from a
value of 0 to a negative will turn the smile into a negative skew with ITM
calls being the most expensive and OTM calls the cheapest and the opposite
for positive . The interpretation of a symmetric smile for a given maturity
is that the market is expecting the underlying to be equally volatile to both
upward and downward movements. In the case of negative the market is
expecting the underlying to be more volatile to a downward movement than
an upward movement for a given maturity and opposite in case of positive .
The former case is often related to stock markets (Hull, 2008) where the the
negative skew can be explained by the Leverage Eect, where higher volatility
on the downside is explained by the decrease in asset value relative to liabilities
making the company/index more risky.
: The vol. of vol. parameter commands the curvature of the smile/skew.
Setting = 0 would yield a constant curve. Higher introduces more convex-
ity to the curve and depending on the value of , the curve would be a steeper
smile tilted to the left in case of positive and titled to the right i case of
negative rho.
v
0
, and : The initial variance, long run variance and speed of mean reversion
decides the term structure of the ATM volatility guarding the level of the
surface.
In an attempt to exemplify the above qualitative eects of changes in the pa-
rameters by looking at Figures 9.6 - 9.21 and comparing the shapes of the imp. vol.
surfaces to the specic values of the parameters in Tables 9.3 - 9.4 we nd that the
changes in values of , v
0
and are the most meaningful and the values of and
to less extend.
The levels of the surfaces are indeed governed by v
0
and with increasing values
considering the EURUSD and unchanged values for the USDJPY. These changes
reect the upward movement in the level of the EURUSD surface, while reecting
an unchanged level in the USDJPY surface. Next the increases in the negative
values of seem reasonable for both EURUSD and USDJPY, where in both cases
the volatility skew seems to be exhibiting more skew in both the short and long term.
It can be noticed that the simulation of the Heston model is doing poorly in
some cases through the examples. The prices and the imp. vols. should indeed
resemble the results calculated from the Heston closed form solution. The problem
seems to narrow down to the 1Y in-the-money prices, where it can be observed that
the imp. vol. retrieved from the simulated solution is too low compared to the imp.
47
vols. from the closed form solution. Although this is not a satisfying result we point
to the fact that it is still a better t to the actual market prices observed than that
achieved by the BS model.
48
10
Empirical study on the hedging
performances
We want to set up a study on the hedging performance of a dynamic BS Delta
hedge with updating implied volatility on each rebalancing point as described in
(Castagna, 2010, p. 63-67). This strategy is carried out on a number of shorted call
options. This way of hedging is an advanced method when compared to a static
BS Delta hedge and a dynamic BS Delta hedge with a constant implied volatility
as described in (Castagna, 2010, p. 61-63). The price of the option is continuously
(approximated by a discretization scheme) marked to market throughout its life up
until expiration. This would in reality imply that the option is actually quoted in
the market. But in this study we instead calculate the price from the Heston model
calibrated to the market prices. From this price we then retrieve the BS implied
volatility to use in the calculation of Delta at the given rebalancing point.
10.1 Size of the study
The study covers a range of options in the maturity dimension from short term, 1M,
to longer term, 1Y, options and in the strike dimension from ITM to OTM options.
In particular a number of 9 options is sold every day at 3 dierent strike levels in
each of the 3 maturities 1M, 6M and 1Y. Considering 371 trading days this gives a
total of 2.142 options. A more detailed picture of the options under investigation is
depicted in Table 10.2.
49
Table 10.1: Number of options under investigation
ITM (ATM) OTM Total
1M 350 350 350 1050
6M 245 245 245 735
1Y 119 119 119 357
Total 714 714 714 2.142
Table 10.2: Number of option expirations in quarterly periods
1M 6M 1Y Total
2010 Q1 120 0 0 120
2010 Q2 189 0 0 189
2010 Q3 192 186 0 378
2010 Q4 192 192 0 384
2011 Q1 186 186 186 558
2011 Q2 171 171 171 513
Total 1050 735 357 2.142
Less options with maturity 1Y is considered compared to maturity 1M, because we
need the options to expire within the 371 trading days in order to calculate the
empirical hedge performance. Therefore we short the last 1Y options on the 119.
trading day with expiration 252 days later on trading day 371. Equivalently, the
last 6M options is sold on trading day 245 expiring 126 days later and 1M options
on trading day 350 expiring 21 days later.
10.2 Strike levels
The strike levels are decided from Equation (10.1), which gives strike points in the
range of 1 standard deviation relative to the forward at-the-money point.
K
i,j
= S
0
exp
_
(r
d
i
r
f
i
)T
i
+
atm
(T
i
)
_
T
i
(1 + j)
_
, j = 0, 1, 2. (10.1)
On each j = 0, j = 1 and j = 2 we get a number of 714 ITM, (ATM) and OTM
options. Deciding the strike points in this way gives us a comparable moneyness
level between the dierent maturities. Alternatively, common strike levels used for
all three maturities would generally lead to very dierent moneyness levels.
This way of generating the strike prices ultimately leads to the average Delta
levels in Table 10.3 for the EURUSD and Table 10.4 for the USDJPY.
50
Table 10.3: Delta level on average of shorted EURUSD call options at initiation
ITM (ATM) OTM
1M 0.831 0.507 0.177
6M 0.853 0.516 0.164
1Y 0.857 0.520 0.161
calculated from Heston prices
Table 10.4: Delta level on average of shorted USDJPY call options at initiation
ITM (ATM) OTM
1M 0.823 0.506 0.156
6M 0.824 0.517 0.158
1Y 0.833 0.506 0.156
calculated from Heston prices
10.3 The hedging portfolio
We take a short position in a call option contract at time t
0
which expires at t
n
= T.
The price of this option is C
0
at initiation and at expiry the option value will be
C
T
. The dynamic delta hedge of this option consists of n discrete steps, where each
step size in this study has a length of 1 trading day. More specically one might
think of the rebalancing points as the end of each trading day as the option data
that we have calibrated the Heston model to consists of ending quotes. In notation
we have the daily rebalancing points {t
0
= 0, t
1
, ....., t
n
= T}. As an example, for
the maturity 1M with a maturity of 21 trading days this gives us 22 rebalancing
points where the rst rebalancing point is on the initial day where the option is sold
followed by 20 intermediate rebalancing points and ending with the closing of the
position as the nal rebalancing point. The action on the nal rebalancing point
where we close down the hedge position consists of either two things:
Option expires ITM: Considering the Delta position that we already hold in the
underlying, we buy the rest up until we hold a notional of 1 of the underlying.
Option expires OTM: Considering the Delta position that we already hold in
the underlying, we sell this to hold a notional of 0 of the underlying.
The general change in the portfolio value,
t
i
with a short position in one call
option contract and a long position in
t
i
units of the underlying currency, where

t
i
is calculated from Equation (7.7), is given by

t
i
=
t
i
(S
t
i
+1
S
t
) + (t
i+1
t
i
)
_
r
d
t
i
(
t
i

t
i
S
t
i
) + r
f
t
i

t
i
S
t
i
_
(10.2)
51
for i = 0, 1, 2, 3, ....., n 1 with an initial portfolio value at i = 0 of
0
= C
0
. In
the calculation of the initial call option contract the interest rates, r
d
t
i
and r
f
t
i
is
the prevailing interest rates at that point in time r
d
t
i
= r
d
(t
i
, T) and r
f
t
i
= r
f
(t
i
, T).
Whereas for the rest of the rebalancing points throughout the life of the given op-
tion starting at i = 1 a simplied interest rate setting is adopted. Here we simply
calculate and use the mean of the interest rate over the period of the option life,
r
d
t
i
=
1
n

n
i=1
r
d
t
i
and equivalently for the foreign interest rate, r
d
t
i
=
1
n

n
i=1
r
f
t
i
. As
an example considering the EURUSD pair
From Equation (10.2) we have that the change in the portfolio value is aected
by three elements. First, the change in the underlying FX rate times the position
in the underlying decided on the rebalancing point at the end of the previous day,

t
i
(S
t
i
+1
S
t
). Second, the nancing of the net portfolio value (stated as account
in the code), r
d
t
i
(
t
i

t
i
S
t
i
). And third, the interest rate received on the position
in the underlying decided on the rebalancing point at the end of the previous day,
r
f
t
i

t
i
S
t
i
.
We dene the hedging cost to be C
T

T
, where the cost is positive if we
have a loss on the hedging strategy and negative if we gain a prot on the hedging
strategy. Furthermore we dene the hedging error as a percentage of the initial spot
rate,
C
T

T
S
0
like it is done in (Thul, 2010). As the ultimate hedging performance
measure we look at the standard deviation of the overall hedging error.
In order to get a sense of the total risk involved in shorting the options, we also
compute an unhedged position. The hedging error here is computed as a percentage
of the initial spot rate, the same way as in the hedging experiments. The hedging
portfolio in this case consists of the compounded sales price of the option. The hedge
portfolio earns the return of the domestic interest rate throughout the maturity of
the option sold. The hedging error is then calculated at expiration of the option as
in calculated as the dierence between the payo of the option and the compounded
initial option price. The initial option price is calculated from the Heston model.
10.4 Results
Looking at the development in the underlying of the option contracts there is a
clear dierence between the development in the EURUSD compared to the USDJPY.
They develop almost exactly opposite over the period considered in the study. From
Figures 10.1 - 10.2 we observe that the EURUSD is decreasing in the beginning and
then increasing over the vast majority of the period, while the USDJPY is increasing
52
a bit in the beginning and then decreasing over the vast majority of the period.
This observation is directly tractable in Tables 10.5 - 10.6 where the number of
option expiring in-the-money is recorded. There is a lot more EURUSD call options
expiring in-the-money than USDJPY expiring in-the-money as a natural result of
the respective developments in the underlying.
!"!#
!"%#
!"&#
!"'#
!"(#
!")#
! &* +( !!% !', !*) %%& %)# %,+ &&' &+!
!
"
#
$

&
'
$
(

)&'*+,- *'.
/010!2
!"
!$
%"
%$
&"
&$
'""
' (% !$ '') '*& '%+ ))( )+" )&! ((* (!'
!
"
#
$

&
'
$
(

)&'*+,- *'.
/!0123
Figure 10.1: Development in EURUSD
spot rate
Figure 10.2: Development in USDJPY
spot rate
Table 10.5: Number of EURUSD call options expiring in-the-money
initiated at ITM (ATM) OTM Total
1M 282 / 81% 176 / 50% 79 / 23% 537 / 51%
6M 229 / 93% 189 / 77% 14 / 6% 432 / 59%
1Y 119 / 100% 91 / 76% 20 / 19% 230 / 64%
Total 630 / 88% 456 / 64% 113 / 16% 1199 / 56%
Table 10.6: Number of USDJPY call options expiring in-the-money
initiated at ITM (ATM) OTM Total
1M 293 / 84% 129 / 37% 23 / 7% 445 / 42%
6M 205 / 84% 22 / 9% 0 / 0% 227 / 31%
1Y 104 / 87% 0 / 0% 0 / 0% 104 / 29%
Total 602 / 84% 151 / 21% 23 / 3% 776 / 36%
The ultimate results of the study are presented in Table 10.7, where the mean
hedging errors and standard deviations under the two dierent model assumptions
are recorded together with the case of the unhedged expirement. As long as the
mean hedging errors are not too far from each other we do not pay special attention
to this measure when evaluating the hedging performances. The overall conclu-
sion in both the case of the EURUSD and USDJPY is that the BS Delta hedge
is performing the best under the assumption of Heston underlying dynamics. The
standard deviation is the lowest in the Heston case followed by the BS model and
highest in the unhedged scenario in both FX pair examples. Furthermore we ob-
serve that the mean hedging error in all three experiments is positive in the case
53
of the EURUSD and negative in the case of the USDJPY. This result is in accor-
dance with the results in (Thul, 2010) where they nd that the hedging errors are
correlated with the underlying returns. Likewise we record a positive hedging error
on call options written on the EURUSD, where we can observe a positive return
development in the period where most of the options expire and visa versa in the
case of the USDJPY.
We do want to draw attention to the fact that the results are obtained based on
a relatively small number of simulations due to scarce supply of computer power.
We have tried to make the estimations comparable by using the same number of
simulations for both the Heston and the BS model. The study could be optimized
and made more accurate by increasing the number of simulations. In this case we
would expect the hedging performances to improve.
When it comes to the explanation of the size of the standard deviation from
which we evaluate the hedging performance one could look at the calibration per-
formances. It seems though that we are not able to conclude anything from this
perspective. The BS model calibration performance was better to the EURUSD
prices in a relative comparison with the Heston t suggesting a hedging perfor-
mance closer to the Heston models. This is not the case in this example.
A theoretical fact is that the cost of rebalancing the Delta are path-dependant
and tied to the option Gamma. The Gamma is strictly positive in case of plain
vanilla options meaning that an increase in the underlying always increases Delta.
In the case of the EURUSD where the spot is generally increasing we incur bigger
losses as we increase the space between rebalancing points. We rebalance only
discretely and therefore incur losses on that account in the case of the EURUSD
and opposite in the case of the USDJPY. So theoretically if we were to increase
the rebalancing points in the period considered we see less of a hedging prot in
the case of the USDJPY. Adjusting the hedge only discretely increases the prots
from rebalancing in the USDJPY but generally comes at the cost of increasing the
standard deviation on the hedging error.
Table 10.7: The mean prot and loss and standard deviation on the hedging error
with Black-Scholes and Heston pricing
EURUSD USDJPY
mean std. dev. mean std. dev.
no hedge +0.81% 3.99% no hedge -2.42% 3.31%
Black-Scholes +0.24% 0.78% Black-Scholes -0.25% 0.61%
Heston +0.20% 0.49% Heston -0.36% 0.46%
54
11
Conclusion
Looking at the distribution of log returns on the EURUSD and USDJPY they exhibit
high peakedness and fat tails. Furthermore, volatility clustering is observed which
is supported by the evidence of autocorrelation in absolute returns. In addition
observations indicate that volatility demonstrates mean reversion. In relation to the
observation of a leptokurtic distribution, nally empirical evidence of heteroscedastic
returns is found, concluding that the volatility of returns is not constant over time.
Thus, we nd empirical evidence against that returns on foreign exchange are
log-normally distributed, which is the assumption in the Black-Scholes option pric-
ing model. The presence of excess kurtosis in log returns suggests that small and
large movements occur more likely compared to normally distributed log returns.
The dierence between the empirically observed dynamics of returns and those as-
sumed in the Black-Scholes model is a result of the severe aw in the model, that
the volatility process is assumed to be constant. This fundamental assumption of
the price process which governs a diusion process for the log asset price that has
stationary and independent normal increments, we nd empirical evidence against.
We convert the quoting of specic FX trading structures published on Bloomberg
into implied volatility quotes on plain vanilla options. Under the assumption of the
put-call parity we then build the volatility surface on call options and calculate the
corresponding option prices. Along this process we retrieve the strike prices from
the Delta moneyness measure by considering the dierent quoting conventions of the
EURUSD and the USDJPY, with the latter being quoted with a premium-included
Delta.
By introduction of a stochastic volatility model like the Heston model, we ob-
serve that the assumptions about the underlying dynamics in this model is much
55
closer to reality. The model performs better in the calibration to observed plain
vanilla options and is able to approximately t the volatility surface, which in both
the EURUSD and the USDJPY exhibits a negative skew increasing with time to
maturity.
A hedging experiment is then set up based on the calibrations of the models. A
dynamic BS Delta hedge with an updating implied volatility simulated in the Heston
model and the Black-Scholes model, respectively, is performed. Given a number of
simulations of the two models we nd that the hedging performance is best when the
pricing is carried out in the Heston model. The standard deviation of the hedging
error is the lowest in the Heston model case on both the EURUSD and the USDJPY.
Both models, although, considerably increase the hedging performance compared to
the alternative of an unhedged position.
Finally we observe that the hedging error is correlated with the underlying re-
turns, which in the case of the EURUSD are positive over the period considered and
negative in case of the USDJPY.
56
Bibliography
Albrecher, H., P. Mayer, W. Schoutens, and J. Tistaert (2005): The
little Heston trap, Wilmott Magazine, pp. 8392.
Anderson, L. (2005): Ecient Simulation of the Heston Stochastic Volatility
Model, SSRN online: http://ssrn.com/abstract=946405.
Bakshi, G., C. Cao, and Z. Chen (1997): Empirical performance of alternative
option pricing models, The Journal of Finance, 52(5), 20032049.
Beneder, R., and M. Elkenbracht-Huizing (2003): Foreign Exchange Op-
tions and the Volatility Smile, Medium Econometrische Toepassingen, 2, 3036.
Black, F. (1976): The pricing of commodity contracts, Journal of Financial
Economics, 3(1-2), 167179.
Black, F., and M. Scholes (1973): The pricing of options and corporate liabil-
ities, Journal of Political Economy, 81(3), 637654.
Bloomberg (2009): FX Volatility Surface Interpolation/Extrapolation FAQ,
Bloomberg.
(2011): FX Volatility Surface Construction, Bloomberg.
Bossens, F., G. Rayee, N. Skantzos, and G. Deelstra (2010): Vanna-Volga
methods applied to FX derivatives: From theory to market practice, Working
Papers CEB.
Breeden, D., and R. Litzenberger (1978): Prices of state-contingent claims
implicit in option prices, The Journal of Business, 51(4), 621651.
Brunner, B., and R. Hafner (2003): Arbitrage-free estimation of the risk-
neutral density from the implied volatility smile, Journal of Computational Fi-
nance, 7(1), 75106.
Carr, P., and D. Madan (1999): Option valuation using the fast Fourier trans-
form, Journal of Computational Finance, 2(4), 6173.
57
Castagna, A. (2010): FX Options and smile risk. Wiley.
Castagna, A., and F. Mercurio (2006): Consistent Pricing of FX Options,
SSRN eLibrary.
Chalamandaris, G., and A. Tsekrekos (2008): Predicting the dynamics of
implied volatility surfaces: A new approach with evidence from OTC currency
options, .
Chinn, M. (2007): Interest rate parity, World Economy.
Cox, J., J. Ingersoll Jr, and S. Ross (1985): A theory of the term structure of
interest rates, Econometrica: Journal of the Econometric Society, pp. 385407.
Elices, A. (2011): Applying hedging strategies to estimate model risk and provi-
sion calculation, Arxiv preprint arXiv 11023534.
Exchange, P. S. (2004): A users guide to currency options, Philadelphia Stock
Exchange online: http://www.phlx.com/products/currency/cug.pdf>.
Garman, M. B., and V. S. Kohlhagen (1983): Foreign Currency Option Val-
ues, Journal of International Money and Finance 2, pp. 231237.
Gatheral, J. (2006): The volatility surface: a practitioners guide. Wiley.
Gilli, M., S. Groe, and E. Schumann (2010): Calibrating the NelsonSiegel
Svensson model, Working Papers.
Glasserman, P. (2003): Monte Carlo Methods in Financial Engineering. Springer
Science+ Business Media Inc.
Hamida, S., and R. Cont (2005): Recovering volatility from option prices by
evolutionary optimization, Journal of Computational Finance, 8(4), 4376.
Heston, S. (1993): A closed-form solution for options with stochastic volatility
with applications to bond and currency options, Review of nancial studies, 6(2),
327.
Hull, C. J. (2008): Options, futures and other Derivatives. Prentice Hall.
Jackwerth, J. (1999): Option-implied risk-neutral distributions and implied bi-
nomial trees: A literature review, Journal of Derivatives, 7(2), 6682.
Janek, A., T. Kluge, R. Weron, and U. Wystup (2010): FX Smile in the
Heston Model, Arxiv preprint arXiv:1010.1617.
58
Jarque, C. M., and A. K. Bera (1987): A test for normality of observations
and regression residuals, International Statistical Review, 55(2), 163172.
Malz, A. (1997): Estimating the probability distribution of the future exchange
rate from option prices, The Journal of Derivatives, 5(2), 1836.
Merton, R. C. (1973): Theory of rational option pricing, Bell Journal of Eco-
nomics and Management Science, 4(1), 141183.
Mikhailov, S., and U. Nogel (2003): Hestons stochastic volatility model
implementation, calibration and some extensions, Wilmott magazine, 4, 7479.
Moodley, N. (2005): The heston model: A practical approach with matlab code,
Faculty of Science, University of the Witwatersrand, South Africa.
Oldfield, G. S., and R. J. Messina (1977): Forward exchange price determi-
nation in continuous time, The Journal of Financial and Quantitative Analysis,
12(3), 473479.
Reiswich, D., and U. Wystrup (2010): FX Volatility Smile Construction,
TBA.
Rouah, F. (2011): Three derivations of the Black-Scholes PDE, Fabrice Rouah
online: http://www.frouah.com.
Thul, M. (2010): Pricing and Hedging of Equity Derivatives under
Volatility Smile consistent Underlying Dynamics, Matthias Thul online:
http://www.matthiasthul.com.
59
A
Retrieving the strike price
corresponding to a premium
included Delta
60
a
61
a
62
B
Building the market implied
volatility surface
63
a
64
a
65
a
66
a
67
a
68
a
69
a
70
a
71
a
72
a
73
a
74
a
75
C
Calibration of the Heston model
76
a
77
a
78
a
79
a
80
a
81
D
Calibration of the Black-Scholes
model
82
a
83
a
84
E
Simulation of the Heston model
85
a
86
a
87
a
88
F
Simulation of the Black-Scholes
model
89
a
90
a
91
G
No hedge
92
a
93
a
94
a
95
a
96
H
Dynamic BS Delta Hedge with
updating imp. vol. from the
Heston model
97
a
98
a
99
a
100
a
101
a
102
a
103
a
104
a
105
a
106
a
107
a
108
I
Dynamic BS Delta Hedge with
updating imp. vol. from the
Black-Scholes model
109

You might also like