Академический Документы
Профессиональный Документы
Культура Документы
0
0S
=
0
0S
0
0
=
1
S
0
0
0
2
0S
2
=
0
0S
_
1
S
0
0
_
=
1
S
0
0S
_
0
0
_
1
S
2
0
0
=
1
S
0
0S
0
0
_
0
0
_
1
S
2
0
0
=
1
S
2
0
2
0
2
1
S
2
0
0
=
1
S
2
_
0
2
0
2
0
0
_
Black Scholes Model
Page 29
Now the BlackScholes equation can be written under this transforma-
tion as
0l
0t
=
1
2
o
2
S
2
1
S
2
_
0
2
0
2
0
0
_
l +vS
1
S
0
0
l
which simplies to
0l
0t
=
1
2
o
2
0
2
l
0
2
+
_
v
1
2
o
_
0l
0
We need to eliminate the rst order derivative term in .
Final Step
Perform a translation of the co-ordinate system
a = +
_
v
1
2
o
2
_
t
Black Scholes Model
Page 30
So we are transforming from (, t) to (a, t) . So apply chain rule I
0
0t
=
0a
0t
0
0a
+
0t
0t
0
0t
=
0
0t
+
_
v
1
2
o
2
_
0
0a
0
0
=
0a
0
0
0a
= 1.
0
0a
==
0
2
0
2
=
0
2
0a
2
So l = W(a, t).
0l
0t
=
1
2
o
2
0
2
l
0
2
+
_
v
1
2
o
_
0l
0
becomes
_
0
0t
+
_
v
1
2
o
2
_
0
0a
_
W =
1
2
o
2
0
2
W
0
2
+
_
v
1
2
o
_
0W
0
After this change of variables the BSE becomes
0W
0t
=
1
2
o
2
0
2
W
0a
2
. (1)
To summarize the steps taken to get this 1D heat equation:,
Black Scholes Model
Page 31
\ (S, t) =
c
v(Tt)
l(S, t) = c
vt
l(S, T t) = c
vt
l(c
, T t)
= c
vt
l
_
c
a
_
v
1
2
o
2
_
t
, T t
_
= c
vt
W(a, t).
So we will start by solving for W(a, t). The equation for this function
is solved using the similarity reduction method, for the fundamental
solution W
)
(a, t; a
t
). This is all familiar methodology. We dene
W
)
(a, t; a
t
) = t
c
)
_
(a a
t
)
t
o
_
,
where a
t
is an arbitrary constant, and the parameters c and o are con-
stant, to be chosen shortly. We choose
(aa
t
)
t
o
because it is a constant
coecient problem.
Black Scholes Model
Page 32
Note that the unknown function depends on only one variable
j = (a a
t
)t
o
Again we use a combination of product and chain rule to write the PDE
in terms of an ODE:
W
)
(a, t; a
t
) = t
c
) (j) ; j = (a a
t
)t
o
So
oj
ot
= ot
o1
(a a
t
);
oj
oa
= t
o
Black Scholes Model
Page 33
0W
0t
= t
c
0
ot
) (j) +ct
c1
) (j)
= t
c
o)
oj
oj
ot
+ct
c1
) (j)
= t
c1o
o
o)
oj
.
_
a a
t
_
+ct
c1
) (j)
= t
c1
_
oj
o)
oj
+c) (j)
_
Black Scholes Model
Page 34
0W
0a
= t
c
0
oa
) (j)
= t
c
o)
oj
oj
oa
= t
c
t
o
o)
oj
= t
co
o)
oj
0
2
W
0a
2
= t
co
0
oa
_
o)
oj
_
= t
co
o
oj
oj
oa
_
o)
oj
_
= t
co
t
o
o
2
)
oj
2
= t
c2o
o
2
)
oj
2
So
0W
0t
= t
c1
_
oj
o)
oj
+c) (j)
_
(2)
Black Scholes Model
Page 35
0
2
W
0a
2
= t
c2o
o
2
)
oj
2
(3)
Substituting (2) , (3) into (1) gives the 2nd order equation
t
c1
_
c) oj
o)
oj
_
=
1
2
o
2
t
c2o
o
2
)
oj
2
(4)
We still have a t term in (4) and for similarity reduction we need to
reduce the dimension of the problem. This implies
c 1 = c 2o == o =
1
2
,
to give
_
c)
1
2
j
o)
oj
_
=
1
2
o
2
o
2
)
oj
2
Black Scholes Model
Page 36
With the correct choice of c, o we want
_
o
o
W
)
(a, t; a
t
)oa = 1 Vt
So
_
o
o
W
)
(a, t; a
t
)oa = t
c
_
R
)
_
aa
t
_
t
_
oa
j =
aa
t
_
t
_
toj = oa
So the integral becomes
t
c
_
R
) (j)
_
toj = t
c+12
_
R
) (j) oj = 1
This implies that t
c+12
should equal one, in order for the solution to
be normalised regardless of time. Therefore c = 12.
) (j) becomes our PDF.
Black Scholes Model
Page 37
The function ) now satises
1
2
_
) +j
o)
oj
_
=
1
2
o
2
o
2
)
oj
2
.
where the left hand side can be expressed as an exact derivative
o
oj
(j)) = o
2
o
2
)
oj
2
.
This can be integrated
j) +o
2
o)
oj
=
where the constant = 0 because as j becomes large, both ) (j) and
)t (j) tend to zero.
Black Scholes Model
Page 38
This is variable separable
j) = o
2
o)
oj
_
o)
)
=
1
o
2
_
joj
ln ) =
1
2o
2
j
2
+1
Taking exponentials of both sides gives
) () = C exp
_
j
2
2o
2
_
C is a normalising constant such that
C
_
R
exp
_
j
2
2o
2
_
oj = 1.
Easy to solve by substituting & =
j
_
2o
_
2oo& = oj, and con-
Black Scholes Model
Page 39
verts the integral to
C
_
2o
_
R
c
&
2
o& = 1
C
_
2o
_
= 1
C =
1
_
2o
)(j) =
1
_
2o
c
j
2
2o
2
.
Replacing j gives us the fundamental solution :
W
)
(a, t; a
t
) =
1
_
2t o
c
(aa
t
)
2
2o
2
t
. (5)
This is the probability density function for a Normal random variable
a having mean of a
t
and standard deviation o
_
t. For t ,= 0, W
)
Black Scholes Model
Page 40
represents a series of Gaussian curves. (5) allows us to nd the solution
of the BSE at dierent points (e.g. a
t
= 2; a
t
= 17, etc.).
Properties of The Solution
We have made sure from our solution method that
_
R
W
)
oa = 1
this has been xed. At a
t
= a (exp 0 = 1)
W
)
(a, t; a
t
) =
1
_
2t o
.
Then as t 0 (close to expiration), W
)
o, the Gaussian curve
becomes taller but the area is conned to unity therefore it becomes
slimmer to compensate. As a moves away from a
t
, exp (o) 0.
W
)
(a, t; a
t
) is plotted below for dierent values of t. If t is large then
Black Scholes Model
Page 41
W
)
is at, as t gets smaller W
)
is increasingly peaked, and focused on
a
t
.
-0.2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
-6 -4 -2 0 2 4 6 8
=0.2
=1.0
=5.0 x =x'=1.0
This behaviour of decay away from one point a
t
, unbounded growth at
that point and constant area means that W
)
has turned in to a Dirac
Black Scholes Model
Page 42
delta function c(a
t
a) as t 0.
Dirac delta function
This is written c
_
a a
t
_
= lim
t0
c
_
a a
t
_
, such that
c
_
a a
t
_
=
_
o a = a
t
0 a ,= a
t
_
o
o
c
_
a a
t
_
oa = 1
or
_
o
0
c
_
a a
t
_
oa = 1
If j (a) is a continuous function then
_
o
o
j (a) c
_
a a
t
_
oa = j
_
a
t
_
Black Scholes Model
Page 43
So if we take a delta function and multiply it by any other function -
and calculate the area under this product - this is simply the function
j (a) evaluated at the point a = a
t
. What is happening here?
The delta function picks out the value of the function at which it is
singular (in this case a
t
). All other points are irrelevant because we are
multiplying by zero.
In the limit as t 0 the function W
)
becomes a delta function at
a = a
t
. This means that
lim
t0
1
o
_
2t
_
o
o
c
(a
t
a)
2
2o
2
t
j(a
t
)oa
t
= j(a).
Here we have swapped a and a
t
- it makes no dierence due to the
(a
t
a)
2
term hence either can be the spatial variable.
Black Scholes Model
Page 44
So
1
o
_
2t
c
(a
t
a)
2
2o
2
t
is a delta function and j(a
t
) will be replaced by the payo function.
The term above is also an example of a Greens function, which allows
us to write down the general solution of the BSE in integral form.
So as we get closer to expiration, i.e. t 0, the delta function picks
out the value of j(a
t
) at which a
t
= a
Now introduce the payo at t = T (t = 0):
\ (S, T) = Payo(S).
Black Scholes Model
Page 45
Recall a = +
_
v
1
2
o
2
_
t, so at expiry t = 0 == a = = log S.
Hence S = c
a
to give
W(a, 0) = Payo(c
a
).
The solution of this for t 0 is
W(a, t) =
_
o
o
W
)
(a, t; a
t
)Payo(c
a
t
) oa
t
.
We have converted the backward BSE to the Forward Equation. Look
at
Payo(c
a
t
) oa
t
.
Black Scholes Model
Page 46
We know
a
t
= log S
t
==oa
t
=
oS
t
S
t
and
c
a
t
= S
t
therefore Payo(c
a
t
).oa
t
becomes
Payo(S
t
).
oS
t
S
t
This result is important. As log S does not exist in the negative plane
the integral goes from 0 to innity, with the lower limit acting as an
asymptote.
Lets start unravelling some of the early steps and transformations. Re-
turning to our Greens function
1
o
_
2(Tt)
c
(a
t
a)
2
2o
2
(Tt)
=
Black Scholes Model
Page 47
1
o
_
2(Tt)
c
_
_
_
_
_
1
2o
2
(Tt)
.
_
_
_
_
_
log S +
_
v
1
2
o
2
__
(T t)
. .
a
log S
t
. .
a
t
_
_
_
_
2
_
_
_
_
_
=
1
o
_
2(Tt)
c
_
1
2o
2
(Tt)
.
__
log
S
S
t
+
_
v
1
2
o
2
__
(Tt)
_
2
_
So putting this together with the Payo function as an integrand we
have
1
o
_
2(Tt)
_
o
0
c
_
1
2o
2
(Tt)
.
__
log
S
S
t
+
_
v
1
2
o
2
__
(Tt)
_
2
_
Payo(S
t
).
oS
t
S
t
\ (S, t) =
c
v(Tt)
o
_
2(T t)
_
log(SS
t
)+
_
v
1
2
o
2
_
(Tt)
_
2
/2o
2
(Tt)
Payo(S
t
)
oS
t
S
t
. (6)
This expression works because the equation is linear - so we just need to
specify the payo condition. It can be applied to any European option
on a single lognormal underlying asset.
Equation (6) gives us the risk-neutral valuation. c
v(Tt)
present val-
ues to today time t. The integral is the expected value of the payo
with respect to the lognormal transition pdf. The future state is
_
S
t
, T
_
and today is (S, t) . So it represents P
_
(S, t)
_
S
t
, T
_
.
Also note the presence of the risk-free IR v in the pdf. So the expected
payo is as if the underlying evolves according to the risk-neutral random
walk
oS
S
= vot +ooA.
Black Scholes Model
Page 49
The real world drift j is now replaced by the risk-free return v. The
delta hedging has eliminated all the associated risk. This means that if
two investors agree on the volatility they will also agree on the price of
the derivatives even if they disagree on the drift.
This brings us on to the idea of risk-neutrality.
So we can think of the option as discounted expectation of the payo
under the assumption that S follows the risk neutral random walk
\ (S, t) = c
v(Tt)
_
o
0
j
_
S, t; S
t
, T
_
\
_
S
t
, T
_
oS
t
where j
_
S, t; S
t
, T
_
represents the transition density and gives the prob-
ability of going from (S, t) to
_
S
t
, T
_
under
oS
S
= vot +ooA.
Black Scholes Model
Page 50
So clearly we have a denition for
j
_
S, t; S
t
, T
_
=
1
oS
t
_
2(T t)
c
_
log(SS
t
)+
_
v
1
2
o
2
_
(Tt)
_
2
/2o
2
(Tt)
.
Two important points
j
_
S, t; S
t
, T
_
is a Greens for the BSE. As the PDE is linear we can
write the solution down as the integrand consisting of this function
and the nal condition.
The BSE is essentially the backward Kolmogorov equation whose
solution is the transition density
j
_
S, t; S
t
, T
_
with
_
S
t
, T
_
xed
and varying (S, t) ; but with the discounting factor.
Black Scholes Model
Page 51
Formula for a call
The call option has the payo function
Payo(S) = max(S 1, 0).
When S < 1, max(S 1, 0) = 0 therefore
_
o
0
=
_
1
0
+
_
o
1
=
_
o
1
*
_
1
0
0 = 0
Expression (6) can then be written as
c
v(Tt)
o
_
2(Tt)
_
o
1
c
_
log(SS
t
)+
_
v
1
2
o
2
_
(Tt)
_
2
2o
2
(Tt)
(S
t
1)
oS
t
S
t
.
Black Scholes Model
Page 52
Return to the variable a
t
= log S
t
==a
t
= log 1S
t
so we can write
the above integral as
c
v(Tt)
o
_
2(Tt)
_
o
log 1
c
_
a
t
+log S+
_
v
1
2
o
2
_
(Tt)
_
2
2o
2
(Tt)
(c
a
t
1) oa
t
=
c
v(Tt)
o
_
2(Tt)
_
o
log 1
c
_
a
t
+log S+
_
v
1
2
o
2
_
(Tt)
_
2
2o
2
(Tt)
c
a
t
oa
t
1
c
v(Tt)
o
_
2(Tt)
_
o
log 1
c
_
a
t
+log S+
_
v
1
2
o
2
_
(Tt)
_
2
2o
2
(Tt)
oa
t
.
Black Scholes Model
Page 53
Just a couple more steps are required to simplify these messy looking
integrals. Lets look at the second integral
1
c
v(Tt)
o
_
2(Tt)
_
o
log 1
c
1
2
_
a
t
+log S+
_
v
1
2
o
2
_
(Tt)
_
2
o
2
(Tt)
oa
t
use the substitution
& =
_
a
t
+log S+
_
v
1
2
o
2
_
(Tt)
_
o
_
(Tt)
o& =
1
o
_
(Tt)
oa
t
o
_
(T t)o& = oa
t
and the limits:
a
t
= o& = o
& = log 1 & =
_
log 1+log S+
_
v
1
2
o
2
_
(Tt)
_
o
_
(Tt)
Black Scholes Model
Page 54
1
c
v(Tt)
o
_
2(Tt)
_
o
_
log S1+
_
v
1
2
o
2
_
(Tt)
_
o
_
(Tt)
c
1
2
&
2
. o
_
(T t)o&
= 1
c
v(Tt)
_
2
_
o
_
log S1+
_
v
1
2
o
2
_
(Tt)
_
o
_
(Tt)
c
1
2
&
2
. o&
= 1
c
v(Tt)
_
2
_
_
log S1+
_
v
1
2
o
2
_
(Tt)
_
o
_
(Tt)
o
c
1
2
&
2
o&
= 1c
v(Tt) 1
_
2
_
o
2
o
c
1
2
&
2
o&
= 1c
v(Tt)
. (o
2
)
The rst integral requires similar treatment however before we do that
we complete the square on the exponent. The integrand is
= c
_
a
t
+log S+
_
v
1
2
o
2
_
(Tt)
_
2
2o
2
(Tt)
c
a
t
v(Tt)
Black Scholes Model
Page 55
Now just work on the exponent, and put t = T t temporarily to
simplify working
_
a
t
+log S+
_
v
1
2
o
2
_
t
_
2
2o
2
t
+a
t
vt
=
_
a
t
+log S+
_
v
1
2
o
2
_
t
_
2
+2(a
t
vt)o
2
t
2o
2
t
=
_
_
a
t
+log S+
_
v
1
2
o
2
_
t
_
2
2(a
t
vt)o
2
t
_
2o
2
t
=
1
2
_
_
a
t
+log S+vt
1
2
o
2
t
_
2
2(a
t
vt)o
2
t
_
o
2
t
Black Scholes Model
Page 56
Now expand the bracket in the numerator
_
a
t
2
+ log
2
S +v
2
t
2
+
1
4
o
2
t
2
2a
t
log S 2a
t
vt +a
t
o
2
t + 2vt log S
o
2
t log S vo
2
t
2
_
2a
t
o
2
t + 2vo
2
t
2
_
a
t
2
+ log
2
S +v
2
t
2
+
1
4
o
2
t
2
2a
t
log S 2a
t
vt a
t
o
2
t + 2vt log S
o
2
t log S +vo
2
t
2
_
now complete the square
=
_
a
t
+ log S +vt +
1
2
o
2
t
_
2
2o
2
t log S
=
_
a
t
+ log S +
_
v +
1
2
o
2
_
t
_
2
2o
2
t log S
Black Scholes Model
Page 57
Lets return to the integral
1
o
_
2t)
_
o
log 1
c
1
2o
2
t
_
_
a
t
+log S+
_
v+
1
2
o
2
_
t
_
2
2o
2
t log S
_
oa
t
=
S
o
_
2(Tt)
_
o
log 1
c
1
2o
2
(Tt)
_
_
a
t
+log S+
_
v+
1
2
o
2
_
(Tt)
_
2
_
oa
t
and as before use a similar substitution
=
_
a
t
+log S+
_
v+
1
2
o
2
_
(Tt)
_
o
_
(Tt)
o =
1
o
_
(Tt)
oa
t
o
_
(T t)o = oa
t
and the limits as before:
a
t
= o& = o
a
t
= log 1 & =
_
log 1+log S+
_
v+
1
2
o
2
_
(Tt)
_
o
_
(Tt)
.
Black Scholes Model
Page 58
Following the earlier working reduces this to
S
1
_
2
_
_
log S1+
_
v+
1
2
o
2
_
(Tt)
_
o
_
(Tt)
o
c
1
2
2
o
= S
1
_
2
_
o
1
o
c
1
2
2
o
= S. (o
1
)
Thus the option price can be written as two separate terms involving
the cumulative distribution function for a Normal distribution:
\ (S, t) = S.(o
1
) 1c
v(Tt)
.(o
2
)
where
o
1
=
log(S1) + (v +
1
2
o
2
)(T t)
o
_
T t
and
Black Scholes Model
Page 59
o
2
=
log(S1) + (v
1
2
o
2
)(T t)
o
_
T t
= o
1
o
_
T t.
.(a) =
1
_
2
_
a
o
c
1
2
2
o.
Black Scholes Model
Page 60
The diagrams above show
a) The value of a call option as a function of the underlying at a xed
time prior to expiry
b) The value of a call option as a function of asset and time
Black Scholes Model
Page 61
Observations:
Call values decrease as the strike increases
Call prices decrease as we get closer to expiry (T t) 0.
Call prices increase with volatility
Call prices increase with interest rates.
Black Scholes Model
Page 62
When there is a continuous dividend yield 1 or the option is on a
currency which receives interest at the foreign rate (replace 1 by v
)
),
then the call option simply becomes
C (S, t) = Sc
1(Tt)
.(o
1
) 1c
v(Tt)
.(o
2
)
where
o
1
=
log(S1) + (v 1 +
1
2
o
2
)(T t)
o
_
T t
and
o
2
=
log(S1) + (v 1
1
2
o
2
)(T t)
o
_
T t
= o
1
o
_
T t.
Black Scholes Model
Page 63
At-The-Money-Forward-Options: A nice approximation
Within the FX world At-The-Money-Forward (ATMF) options are the
most heavily traded. When an option is struck ATMF, it means that the
strike 1 = Sc
(v1)t
, where we use the earlier denition of t = T t.
This is because the call and put are equal. The put-call parity when
there is a dividend yield is
C (S, t) 1 (S, t) = Sc
1t
1c
vt
which for ATMF is
Sc
1t
= 1c
vt
.
There exists a very nice approximation for ATMF options near expiry.
Black Scholes Model
Page 64
Begin by writing the call option formula
C (S, t) = Sc
1t
.(o
1
) Sc
(v1)t
c
vt
.(o
2
)
= Sc
1t
(.(o
1
) .(o
2
))
Now simplify o
1
and o
2
o
1
=
log
_
SSc
(v1)t
_
+
_
v 1 +
1
2
o
2
_
t
o
_
t
=
log c
(v1)t
+
_
v 1 +
1
2
o
2
_
t
o
_
t
=
(v 1) t +
_
v 1 +
1
2
o
2
_
t
o
_
t
=
1
2
o
2
t
o
_
t
=
1
2
o
_
t
Similar working shows o
2
=
1
2
o
_
t.
Black Scholes Model
Page 65
Returning to the earlier denition of the option price
C (S, t) = Sc
1t
_
.
_
1
2
o
_
t
_
.
_
1
2
o
_
t
__
Consider the CDF for a variable a, i.e. . (a) . This can be approximated
due to Kendall and Stuart (1943)
. (a) =
1
2
+
1
_
2
_
a
a
3
6
+
a
5
40
+O
_
a
7
_
_
.
If a is small then to leading order this becomes . (a) ~
1
2
+
1
_
2
a.
So if we are close to expiry, then t (= T t) is small hence
.
_
1
2
o
_
t
_
~
1
2
1
_
2
_
1
2
o
_
t
_
_
.
_
1
2
o
_
t
_
.
_
1
2
o
_
t
__
=
1
_
2
o
_
t ~ 0.4o
_
t
Black Scholes Model
Page 66
Putting this altogether gives
C (S, t) ~ 0.4Sc
1(Tt)
o
_
T t
Black Scholes Model
Page 67
Formula for a put
The put option has payo
Payo(S) = max(1 S, 0).
A similar working as in the case of a call yields
\ (S, t) = S.(o
1
) +1c
v(Tt)
.(o
2
),
with the same o
1
and o
2
. Naturally the more sensible approach is to
exploit the put-call parity. If the price of a call and put are denoted in
turn by C (S, t) and 1 (S, t)
C 1 = S 1c
v(Tt)
Black Scholes Model
Page 68
hence rearranging, using the formula for a call together with . (a) +
. (a) = 1, gives
1 = C S +1c
v(Tt)
= S.(o
1
) 1c
v(Tt)
.(o
2
) S +1c
v(Tt)
= S(.(o
1
) 1)
. .
=.(o
1
)
+1c
v(Tt)
(1 .(o
2
))
. .
=.(o
2
)
= S. (o
1
) +1c
v(Tt)
. (o
2
)
Black Scholes Model
Page 69
The diagrams above show
a) The value of a put option as a function of the underlying at a xed
time prior to expiry
b) The value of a put option as a function of asset and time
Black Scholes Model
Page 70
Observations:
Put values increase as the strike increases
Put prices increase as we get closer to expiry (T t) 0.
Put prices increase with volatility
Put prices increase as interest rates decrease.
Black Scholes Model
Page 71
Binary Options
Also known as digital options. These have discontinuous payos. Two
general types: cash-or-nothing or asset-or-nothing options.
In the rst type, a xed amount of cash is paid at expiry if option is
in-the-money, whilst the second pays out the value of the underlying
asset. The payo is dened in terms of the Heaviside function
1(a) =
_
1 a 0
0 a < 0
which for a binary call option becomes
Payo (S) = 1(S) =
_
1 S (T) 1
0 otherwise
So 1 takes the value one when it has a positive argument and zero
otherwise. So if the option
Black Scholes Model
Page 72
The diagram shows the value of a binary call sometime before expiration.
What is happening here?
Each part of the BSE plays a role here.
Black Scholes Model
Page 73
v\ : Present valuing, has the aect of discounting
1
2
o
2
S
2
0
2
\
0S
2
: diusion rounds o the corners
+vS
0\
0S
: convection shifts the prole to the left
Pricing Formulas:
Cash-or-nothing
C = c
v
o
t
. (o
2
) FOR call/DOM put
1 = c
v
o
t
. (o
2
) FOR put/DOM call
Asset-or-nothing
C = Sc
v
)
t
. (o
1
) FOR call/DOM put
1 = Sc
v
)
t
. (o
1
) FOR put/DOM call
Black Scholes Model
Page 74
The greeks
We now examine the sensitivity of an option price to the input vari-
ables/parameters.
The Greeks are forms of measurement on options that express the
change of the option price when some parameter changes given every
other parameter stays the same. This is an essential form of risk man-
agement carried out by all option traders. The next table denes some
of the basic greeks.
Black Scholes Model
Page 75
greek symbol Measures change in
delta =
0\
0S
option price change when underlying price
increases by1
theta =
0\
0t
option price when time to expiry
decreases by 1 day
gamma =
0
0S
delta when the stock price
increases by 1
vega
0\
0o
option price when volatility
increases by1% (100 basis points)
rho j =
0\
0v
option price when interest rate
increases by 1% (100 basis points)
psi =
0\
01
option price when dividend yield
increases by 1%
Black Scholes Model
Page 76
The greeks above are all rst order derivatives with the exception of
gamma which is
0
2
\
0S
2
and from the list is the only sensitivity that does not
measure a change in the option price change, but rather, it measures the
change in delta. Theta is the only Greek that is in the negative domain as
it measures decreases in time. There is no greek letter assigned to vega.
vega, j and all measure one percent increases (100 basis points),
such as the risk-free rate increasing from 3.5% to 6% (an example of
j).
Black Scholes Model
Page 77
Delta
The graph illustrates the behaviour of both call and put option deltas
for Europeans and Binaries as they shift from being OTM to ATM and
nally ITM. Note that calls and puts have opposite deltas - call options
are positive and put options are negative. The binary deltas have been
rescaled so they can be observed on the same plot.
Black Scholes Model
Page 78
Here \ can be the value of a single contract or of a whole portfolio
of contracts. The delta of a portfolio of options is just the sum of the
deltas of all the individual positions.
Option delta is represented as the price change given a 1 point move in
the underlying asset and is usually displayed as a decimal value.
Delta values range between 0 and 1 for call options and 1 to 0 for
put options (which means decimal notation). Some traders refer to the
delta as a whole number between 0 to 100 for call options and 100
to 0 for put options. We can write
=
\ (S +cS, t) \ (S, t)
cS
where cS is the unit move. Since = (S, t) , this means that the
number of assets held must be continuously changed to maintain a delta
neutral position, i.e. = 0. This procedure is called dynamic hedging.
Black Scholes Model
Page 79
Changing the number of assets held requires the continual purchase
and/or sale of the stock. This is called rehedging or rebalancing the
portfolio.
All six of the Greeks are mathematical derivatives of the option price
formula with respect to the given parameters. Gamma, as already men-
tioned, is actually a 2nd derivative of the option price.
Black Scholes Model
Page 80
Gamma
Delta and Gamma are arguably the two most important sensitivities as
they are partial derivatives with respect to the underlying stock. The
change in option price is at the greatest percentage of the option price
when the option is close to a payo of zero. This is when Gamma is at
its highest values. It can be thought of as the acceleration of the option
when the stock changes. This information can be used to predict how
much can be made or lost based on the movement of the underlying
position. Since gamma is the sensitivity of the delta to the underlying it
is a measure of by how much or how often a position must be rehedged
in order to maintain a delta-neutral position.
Black Scholes Model
Page 81
A list of basic greeks was presented in the lecture. More advanced
greeks are discussed in Espens lecture and an exhaustive collection can
be found in his book. Here we use basic dierentiation techniques to
demonstrate the simplicity in obtaining (for example) the delta of a
European Put. The idea is to show how straightforward it actually is
to produce complex looking formulae. We have just written the price of
a put
1 (S, t) = 1c
v(Tt)
. (o
2
) Sc
1(Tt)
. (o
1
)
So we want =
01
0S
.
Useful results:
If .(a) =
1
_
2
_
a
o
c
2
2
o then
o.
oa
=
1
_
2
c
a
2
2
: Leibniz Rule
Black Scholes Model
Page 82
o
1
=
log (S1) +
_
v 1 +
1
2
o
2
_
(T t)
o
_
T t
,
o
2
= o
1
o
_
T t
_
_
==
0 (o
1
)
0S
=
0 (o
2
)
0S
Another result of importance (messy to prove)
Sc
1(Tt)
1
_
2
exp(o
2
1
2) = 1c
v(Tt)
1
_
2
exp(o
2
2
2)
Write
1c
v(Tt)
. (o
2
) (a)
Sc
1(Tt)
. (o
1
) (b)
and
0
0S
(a) = 1c
v(Tt)
0
0S
. (o
2
)
Black Scholes Model
Page 83
now use chain rule
1c
v(Tt)
0
0o
2
. (o
2
)
0 (o
2
)
0S
= 1c
v(Tt)
1
_
2
exp(o
2
2
2)
0 (o
2
)
0S
0
0S
(b) = c
1(Tt)
0
0S
S. (o
1
)
use product rule then chain rule
c
1(Tt)
_
. (o
1
) +S
0
0S
. (o
1
)
_
= c
1(Tt)
_
. (o
1
) +S
0
0o
1
. (o
1
)
0 (o
1
)
0S
_
= c
1(Tt)
_
. (o
1
) S
1
_
2
exp(o
2
1
2)
0 (o
1
)
0S
_
Black Scholes Model
Page 84
So now
=
0
0S
(a)
0
0S
(b)
= 1c
v(Tt)
1
_
2
exp(o
2
2
2)
0 (o
2
)
0S
c
1(Tt)
_
. (o
1
) S
1
_
2
exp(o
2
1
2)
0 (o
1
)
0S
_
= c
1(Tt)
. (o
1
) +
0 (o
1
)
0S
_
Sc
1(Tt)
1
_
2
exp(o
2
1
2) 1c
v(Tt)
1
_
2
exp(o
2
2
2)
_
= c
1(Tt)
. (o
1
) +
0 (o
1
)
0S
(0)
Using
. (a) +. (a) = 1 ==. (a) = 1 . (a)
= c
1(Tt)
(1 . (o
1
))
= c
1(Tt)
(. (o
1
) 1)
Black Scholes Model
Page 85
Basket Options
In reality options are often written on several underlyings. This is an
example of higher dimensional problem and leads onto the idea of multi-
factor models, as there are now more sources of randomness. Consider
the simplest case of a basket option on two stocks S
1
and S
2
oS
i
= j
i
S
i
ot +o
i
S
i
oA, i = 1, 2.
So each asset has its own parameters. oA
1
and oA
2
make this a
two factor model and we will derive an equation for the option price
\ (S
1
, S
2
, t) . The pair of random sources mean that we now require
two assets with which to hedge away our risk. The portfolio becomes
= \ (S, t)
1
S
1
2
S
2
.
As before keeping delta xed across a time step ot gives
o = o\ (S, t)
1
S
1
2
S
2
.
Black Scholes Model
Page 86
We need to consider It for \ (S
1
, S
2
, t)
o =
0\
0t
ot +
1
2
0
2
\
0S
2
1
oS
2
1
+
1
2
0
2
\
0S
2
2
oS
2
2
+
0
2
\
0S
1
0S
2
oS
1
oS
2
_
0\
0S
1
1
_
oS
1
_
0\
0S
2
2
_
oS
2
Note
oS
2
1
= o
2
1
S
2
1
ot; oS
2
2
= o
2
2
S
2
2
ot; oS
1
oS
2
= jo
1
o
2
S
1
S
2
ot
To eliminate risk means we take
1
=
0\
0S
1
,
2
=
0\
0S
2
which gives a risk free portfolio, i.e. no arbitrage, hence
o = vot
Black Scholes Model
Page 87
and the pricing PDE becomes
0\
0t
+
1
2
o
2
1
S
2
1
0
2
\
0S
2
1
+
1
2
o
2
2
S
2
2
0
2
\
0S
2
2
+jo
1
o
2
S
1
S
2
0
2
\
0S
1
0S
2
= v
_
\ S
1
0\
0S
1
S
2
0\
0S
2
_
.
Black Scholes Model
Page 88