Sie sind auf Seite 1von 10

E4706 Foundations of Financial Engineering

Summer 2013
The random variable S
T
is called lognormal, meaning that its log is normal:
logS
T
N(logS
0
+ (r
1
2

2
)T,
2
T).
The density function for Y is
f
Y
(y) =
1

2
e
y
2
/2
.
Remark: We can also compute the pdf of S
T
(note log S
T
here). To do so, we use the fact
that S
T
has the same distribution of
V := exp ( + Z) , Z N(0, 1),
with parameters = log s + (r

2
2
)(T t) and =

T t. Next, we compute the


probability
IP
Q
{S
T
v} = IP
Q
{V v} =
_
log v

_
. (1)
This gives the cdf of S
T
. In turn, we dierentiate w.r.t. v to get the pdf
1
v

_
log v

_
.
Using the lognormal distribution of S
T
, the binomial pricing formula in the continuous-
time limit becomes
V
0
= e
rT
E
Q
_
h(S
0
e
(r
1
2

2
)T+

TY
)
_
=
e
rT

h(S
0
e
(r
1
2

2
)T+

Ty
)e
y
2
/2
dy,
for a general payo function h().
This is the well-known Black-Scholes formula for option pricing. For calls and puts, it is
usually written as an explicit formula in terms of the cumulative normal distribution function
(z) =
1

e
y
2
/2
dy.
For a call option with strike K and maturity T, the Black-Scholes risk-neutral price at time
t, given S
t
= s, is
C(t, s; r, , T, K) = E
Q
{e
r(Tt)
(S
T
K)
+
|S
t
= s} = s(d
1
) e
r(Tt)
K(d
2
), (2)
where
logS
T
N(logS
t
+ (r
1
2

2
)(T t),
2
(T t))
and
d
1
=
log(s/K) + (r+

2
2
)(T t)

(T t)
, d
2
= d
1

T t. (3)
To derive formula (2), we rst break down the expectation into two terms:
E
Q
{e
r(Tt)
(S
T
K)
+
|S
t
= s} = e
r(Tt)
_
E
Q
{S
T
1
{S
T
K}
|S
t
= s} KE
Q
{1
{S
T
K}
|S
t
= s}
_
= e
r(Tt)
_
E
Q
{S
T
1
{S
T
K}
|S
t
= s}
. .
(I)
KP
Q
{S
T
K|S
t
= s}
. .
(II)
_
.
(4)
Then, we apply the normal distribution of log S
T
to compute (I) and (II). Specically, let us
write
log S
T
= log s + (r

2
2
)(T t) +

(T t)Z, for Z N(0, 1).


In turn, the second term can be expressed as
(II) = P
Q
_
log s log K + (r

2
2
)(T t)

(T t)
Z
_
= (d
2
).
To compute the rst term (I), we rst denote = log s + (r

2
2
)(T t) and
2
=
2
(T t)
so that X := log S
T
N(,
2
). In turn, using the normal pdf f
X
, we write
(I) =

e
x
1
{xlog K}
f
X
(x) dx
=


log K
1

2
e
x
e
(x)
2
2
2
dx
= e
+

2
2


log K
1

2
e
(x(+
2
))
2
2
2
dx
= e
+

2
2

_
+
2
log K

_
= se
r(Tt)

_
log s log K + (r +
1
2

2
)(T t)

T t
_
= e
r(Tt)
s(d
1
).
In the last line weve used the denitions of and . As a result, we obtain the rst term of
the Black-Scholes formula (2).
The Black-Scholes formula is a price function in (t, s). In order to measure the sensitivities
of the option price with respect to changes in time and stock price, we consider its partial
derivatives which are commonly referred to as the Greeks. The most common ones are
Theta:
C
t
; Delta:
C
s
, Gamma:

2
C
s
2
.
Some facts that aid in the calculations are
d
1
s
=
d
2
s
=
1
s

T t
and
d
2
t
=
d
1
t
+

2

T t
and
(d
2
) = (d
1

T t) =
1

2
exp
_

1
2
(d
1

T t)
2
_
= (d
1
) exp
_
d
1

T t
1
2

2
(T t)
_
= (d
1
)
s
K
e
r(Tt)
.
Applying chain rule and these facts, the partial derivatives are
=
s(d
1
)
2

T t
re
r(Tt)
K(d
2
),
= (d
1
),
=
(d
1
)
s

T t
.
The rst Greek is negative, meaning that the call option loses value over time (assuming
the stock price is held xed). This is commonly referred to as the time decay associated
with a call. On the other hand, both and are positive. Also, measures the convexity
of the option price with respect to the underlying stock price.
Using the explicit expressions of , and , one can directly verify that the call option
price in (2) satises
C = s +
1
r
_
+
1
2

2
s
2

_
. (5)
In other words, the call option price can be expressed in terms of three Greeks. In particular,
represents the number of shares of stock in the replicating portfolio, so the rst term s
represents the cash amount invested in the stock. The second term represents the money in
the bank. Moreover, notice that (0, 1) and lim
s0
= 0 and lim
s+
= 1. Once the
stock price is very large, it is almost certain that the option will be exercised in the money.
On the other hand, if the stock price is very close to zero, then the option will very likely
end up expiring out of the money, resulting a zero payo. Therefore, it is intuitive that the
(hedge) should change accordingly.
Rearranging equation (5), we can express as
= s
2

2
r(sC). (6)
This gives an interesting interpretation of . To see this, recall that the call price is the same
as its replicating portfolio value, and that the replicating portfolio of a call involves holding
shares of stock and borrowing the amount Cs (at time t when the stock price S
t
= s). The
second term in (6) is simply the interest being paid on the amount borrowed to nance the
purchase of stock shares. Thats why its proportional to the amount borrowed and interest
rate r. As for the other term, remember that describes the change in option value as stock
price is held xed, so imagine that time passes from t to u with S
t
= S
u
(suppose u = t +).
During [t, u], the stock price may uctuate. Since > 0, when the stock price rises, you buy
more stock; but as the stock price falls back, you need to reduce your stock holdings so
you lose money. Similarly, you would also lose money by selling low and buying high when
the stock price rst falls and comes back up. Consequently, the rst term reects that the
replicating strategy loses money at a rate proportional to and to the instantaneous variance
of the stock price
2
s
2
. Intuitively speaking, if the stock is more volatile ( large), your
replicating strategy tends to lose more money. Also, if the price is more convex ( large), you
tend to have to adjust your strategy more signicantly, leading to more leakage.
Rearranging equation (5) again, we get
+
1
2

2
s
2
+ rsrC = 0,
which is indeed the well-known Black-Scholes PDE.
For a put option with strike K and expiration date T, the Black-Scholes price formula is
P(t, s; r, , T, K) = E
Q
{e
r(Tt)
(K S
T
)
+
|S
t
= s} = e
r(Tt)
K(d
2
) s(d
1
). (7)
Remark: Alternatively, we can derive the same put price formula using Put-Call Parity.
Precisely, we have
P(t, s; r, , T, K) = C(t, s; r, , T, K) S + e
r(Tt)
K,
which simplies to (7). This is much less tedious way to obtain the price formula!
We can also compute the Greeks for puts, namely,
=
s(d
1
)
2

T t
+ re
r(Tt)
K(d
2
),
= (d
1
),
=
(d
1
)
s

T t
.
Again, for a put represents the number of shares of stock in the replicating portfolio, so
the rst term s represents the cash amount invested in the stock. Note that is negative
for a put, but positive for a call. Also, notice that the expression for is the same for a call
and a put (with identical strike and maturity).
Interestingly, the put option price function also satises the equation
+
1
2

2
s
2
+ rsrP = 0,
where, of course, , and here are the Greeks associated with the put price P.
Exercise
Verify that the Greeks also satisfy Put-Call Parity. This is no surprise since we can simply
take partial derivative(s) from the Put-Call Parity equation:
C(t, s) P(t, s) = s e
r(Tt)
K.
This also mean that we can obtain the Greeks of a put using those of a call.
Exercise
A long straddle is a combination of a long call with strike K
2
and maturity T and a long
put with strike K
1
and maturity T, with K
1
K
2
. By no-arbitrage principle, the price of a
straddle C
STR
(t, s) is the sum of the corresponding call price and the put price. Whatre the
, , and for a long straddle? Is its positive? What about and ?
Hint: the partial derivative of C
STR
is also a sum of the partial derivatives of the corresponding
call and put.
Exercise
Another important Greek is vega (but denoted by ):
:=
C

= s(d
1
)

T t > 0.
This is the partial derivative of the option price with respect to the volatility parameter
(holding (t, s) xed). In contrast to time t and stock price s, is not even a variable but a
constant parameter of the model. Although this is true, but for traders it is very important
to understand the price sensitivity in the stocks volatility. Note that also varies with time
and stock price.
Compute the vega of a put. Hint: you can use Put-Call Parity. What do you notice?
Exercise
Write vega in terms of .
Exercise
Determine the limit of the Black-Scholes formula for a call and for a put, in each of the
following cases:
s +,
s 0,
+.
Exercise
Show that
d
1
K
=
d
2
K
=
1
K

T t
.
Also, use this result to show that
d
1
K
= e
r(Tt)
(d
2
).
Implied Volatility
One useful way to compare option prices, e.g. across strikes and maturities, is to express
each of them in terms of its implied volatility. Given an observed market price, denoted by
C
obs
(K, T), of a European call option with strike K and maturity T, the implied volatility
I is dened to be the value of the volatility parameter to be plugged into the Black-Scholes
formula (2) to match this price:
C(t, s; r, I, T, K) = C
obs
(K, T) (8)
The Black-Scholes formula is monotonically increasing in :
C

= = s(d
1
)

T t > 0.
Therefore, a unique nonnegative implied volatility I > 0 can be found as long as C
obs
(K, T) >
C(t, s; r, 0, T, K). In view of this one-to-one correspondence, very often traders quote option
prices in terms of implied volatility.
Assume market prices follow Put-Call Parity:
P
obs
= C
obs
s + Ke
r(Tt)
,
where s is the current underlying stock price. Then, the implied volatilities for put and call
options with identical strike and maturity are the same. This follows directly from (8) and
that the Black-Scholes put and call prices also follow Put-Call Parity:
P
obs
= C
obs
S + Ke
r(Tt)
= C(t, s; r, I, T, K) s + Ke
r(Tt)
= P(t, s; r, I, T, K).
Lets consider options with the same expiration date T and hold t, s, r xed. Then, one
can plot the implied volatility Iagainst strike K. One of the most salient feature of the graph
I(K) over K obtained from market option prices is its (symmetric/asymmetric) -shape.
This is called the implied volatility smile/smirk. The implied volatility curve informs traders
that there is a premium charged for out-of-the-money puts and in-the-money calls, i.e. low
strike K, as compared to at-the-money options. To interpret this, we can think as if the
market is averse to large downward stock price movements, and therefore, accounts for it by
adjusting the Black-Scholes price formula.
One may wonder whether there are any bounds on the slope of implied volatility curve
I(K). Recall that (observed/Black-Scholes) call prices must be decreasing in strike K by
no-arbitrage principle. Dierentiating C(t, s; r, I(K), T, K) (holding other parameters xed),
we get
C
obs
K
=
C(t, s; r, I(K), T, K)
K
=
C
K
+
C

I(K)
K
0.
Note that
C
K
(respectively,
C

) is the partial derivative of the Black-Scholes price formula


with respect to its parameter K (respectively ), holding other parameters xed. Rearranging
the above inequality, we obtain the bound on the slope of I(K):
I(K)
K

C/K
C/
.
In contrast, put prices must be increasing in strike K. We do the same calculations for
puts to get the bound:
I(K)
K

P/K
P/
.
Recall that puts and calls with the same K and T also have the same implied volatility I(K).
Hence, we can combine the two bounds and conclude:

2
s

T t
(1 (d
2
))e
r(Tt)+d
2
1
/2

I
K

2
s

T t
(d
2
)e
r(Tt)+d
2
1
/2
,
where d
1
and d
2
are dened in (3) with replaced by I(K). Loosely speaking, the slope of
the implied volatility curve cannot be too negative or too positive, with the bounds given
above.
Exercise
Suppose you observe all the call prices for various strikes with the same maturity T, and the
interest rate is r. Therefore, among many things, you can tell the slope of the price curve at
some strike K

. So, lets say


C
obs
K
(K

) = 0.25. What is the slope of the implied volatility


curve at the same strike K

?
60 70 80 90 100 110 120
0
5
10
15
20
25
30
35
40
45
Stri ke
O
p
t
i
o
n
P
r
i
c
e
0.6 0.5 0.4 0.3 0.2 0.1 0 0.1 0.2
0
5
10
15
20
25
30
35
40
45
Log- Mone y ne ss
O
p
t
i
o
n
P
r
i
c
e
Figure 1: Call option prices plotted against strike K (left) and log-moneyness log(K/S
0
)
(right). In both gures, the left-end corresponds to in-the-money calls (or equivalently, out-
of-the-money puts). As expected, call prices are decreasing in strike or log-moneyness.
0.6 0.5 0.4 0.3 0.2 0.1 0 0.1 0.2
0.2
0.22
0.24
0.26
0.28
0.3
0.32
0.34
0.36
Log- Mone y ne ss
I
m
p
l
i
e
d
V
o
l
a
t
i
l
i
t
y
Figure 2: The implied volatility curve from the option prices in Figure 1. The curve bottoms
approximately at the money (when log-moneyness 0).
Market Implied Risk-Neutral Distribution
In this section, we discuss how market option prices can help determine the risk-neutral
distribution of the underlying stock. To start, recall that the price of a European call on S
with strike K and maturity T is given by
C
0
(K) = e
rT
IE
Q
{(S
T
K)1
S
T
K
} = e
rT


K
(S K) g(S)dS,
where g() is the risk-neutral probability density function of the random stock price S
T
. Weve
written C
0
(K) to highlight it as a function of strike K.
Next, dierentiate once w.r.t. K, we get
C
0
(K)
K
= e
rT


K
g(S)dS.
Dierentiating it again, we get the second-order derivative

2
C
0
(K)
K
2
= e
rT
g(K).
Rearranging, we see that the pdf g at K can be estimated by the second derivative of call
price w.r.t. K, namely,
g(K) = e
rT

2
C
0
(K)
K
2
.
In practice, we can pick three strikes K , K, K + for very small , and then estimate
g(K) using
e
rT
C
0
(K ) 2C
0
(K) + C
0
(K + )

2
.
Exercise
What is

2
P
0
(K)
K
2
, where P
0
(K) is the time-0 put price with strike K.
Exercise
For deep in-the-money calls (and deep out-of-the-money puts), the price as a function of K
is much more linear, as suggested by Figure 1. What does it mean to the pdf g(K) at these
extremely low strikes?
Hedging with Greeks
As we have seen, stock price volatility appears to be better treated as random, rather than
constant in the Black-Scholes model. Indeed, much research has been done to model volatility
as a stochastic process. In this class, however, we focus on the practical problem of using
dierent nancial derivatives to hedge against the risk of changing volatility. This is particular
important for traders who want to to de-sensitize (or immunize) a portfolio (of options, for
example) against changes in volatility, and for the eects of infrequent delta hedging. The
derivatives used are usually vanilla contracts such as put and calls due to their availability
and high liquidity.
Suppose P is the value of a given portfolio. This is likely to be a large combination of many
derivatives positions. Here, for simplicity, we assume that P consists of cash, the underlying
stock S and options all written on S. Then the (P), (P) and (P) of the portfolio are
dened as
(P) =
P
s
, (P) =

2
P
s
2
, (P) =
P

.
Example
For instance, if the portfolio is a combination of n dierent options, each with quantity w
i
(1 i n), then the delta of the portfolio can be written as
(P) =
n

i=1
w
i

i
,
where
i
is the delta of the ith option. Moreover, one can adjust the weights w
i
so that the
delta of the portfolio is zero. In this case, the portfolio is called delta-neutral.
Suppose the portfolio consists of the following options:
A long position in 100,000 calls with strike K
1
and maturity T
1
. The delta of each
option is 0.533.
A short position in 200,000 calls with strike K
2
and maturity T
2
. The delta of each
option is 0.468.
A short position in 50,000 puts with strike K
3
and maturity T
3
. The delta of each option
is 0.508.
Then, the delta of the entire portfolio is
100, 000 0.533 200, 000 0.468 50, 000 (0.508) = 14, 900.
How do we make it delta-neutral ((P) = 0)? We can use the underlying stock S, which
has a per-unit delta one. Therefore, we simply need to add to the portfolio a long position of
14,900 shares of S.
Delta neutrality provides protection against small stock price movement prior to the next
reblancing, but does not protect against larger price movements. This motivates traders to
incorporate options to make a delta-neutral portfolio also gamma-neutral. To see this,
suppose a delta-neutral portfolio has a gamma of 3000. There is a market-traded call C
with = 0.62 and = 1.5. To make the portfolio gamma-neutral, we include a long position
of
3000
1.5
= 2000
units of this call C. However, after this incorporation, the delta of the portoio goes from
zero to
2000 0.62 = 1240,
so its not delta-neutral. We can make the portfolio delta-neutral again by selling exactly
1240 shares of the underlying stock S. Notice the stock S has zero gamma, so selling S will
not aect the gamma of the portfolio while adjusting delta.
On top of delta and gamma, traders are also concerned about the rate of change of portfolio
with respect to the volatility. Also, a portfolio that is both delta-neutral and gamma-neutral
does not imply that it is vega-neutral (vega= 0). To make the portfolio vega-neutral, one
needs to use more traded derivative(s).
Exercise
You have a portfolio P that is neutral but has the following Gamma and Vega:
(P) = 3000, (P) = 4000.
In the market there are two derivatives U
1
and U
2
with the following characteristics:

U
1
0.4 1.5 0.5
U
2
0.8 1 0.4
(a) Explain how you would modify your portfolio to make it , and neutral. Give
the quantities involved.
(Answer: x = 250 and y = 3625, but also sell 1575 shares of underlying stock.)
(b) The numbers U
i
denote the prices of the new derivatives and S is the current stock price:
U
1
= 1.5 U
2
= 1 S = 2.
What is the cost of modifying your portfolio as in part (a)?
(Answer: Cost = 250*1.5 + 3625*1 - 2*1575 = 850.)

Das könnte Ihnen auch gefallen