Topic 1 – Lattice tree methods

1.1 Binomial option pricing models

• Risk neutral valuation principle

• Multiperiod extension
• Early exercise feature and callable feature — dynamic programming
procedure
• Discrete dividend models
• Applications to path dependent options

1.2 Trinomial schemes

• Discounted expectation approach

• Multistate extension – Ritchken-Kamrad’s approach

1.3 Forward shooting grid algorithms (strongly path dependent options)

• Cumulative Parisian feature

• Call options with strike reset feature
• Floating strike arithmetic averaging call
• Alpha-quantile options

1
1.1 Binomial option pricing models

Risk neutral valuation principle

By buying the asset and borrowing cash (in the form of riskless invest-
ment) in appropriate proportions, one can replicate the position of a call.

Under the binomial random walk model, the asset prices after one period
∆t will be either uS or dS with probability q and 1 − q, respectively.

We assume u > 1 > d so that uS and dS represent the up-move and down-
move of the asset price, respectively. The proportional jump parameters
u and d will be related to the asset price dynamics.

2
Let R denote the growth factor of riskless investment over one period so
that $1 invested in a riskless money market account will grow to $R after
one period. In order to avoid riskless arbitrage opportunities, we must
have u > R > d.

Suppose we form a portfolio which consists of α units of asset and cash

amount B in the form of riskless investment (money market account).
After one period of time △t, the value of the portfolio becomes
(
αuS + RM with probability q
αdS + RM with probability 1 − q.

3
The portfolio is used to replicate the long position of a call option on a
non-dividend paying asset.

As there are two possible states of the world: asset price goes up or down,
the call is thus a contingent claim.

Suppose the current time is only one period △t prior to expiration. Let c
denote the current call price, and cu and cd denote the call price after one
period (which is the expiration time in the present context) corresponding
to the up-move and down-move of the asset price, respectively.

4
Let X denote the strike price of the call. The payoff of the call at expiry
is given by
(
cu = max(uS − X, 0) with probability q
cd = max(dS − X, 0) with probability 1 − q.

Evolution of the asset price S and money market account M after one
time period under the binomial model. The risky asset value may either
go up to uS or go down to dS, while the riskless investment amount M
grows to RM .

5
Replicating procedure

The above portfolio containing the risky asset and money market account
is said to replicate the long position of the call if and only if the values of
the portfolio and the call option match for each possible outcome, that
is,
αuS + RM = cu and αdS + RM = cd.
Solving the equations, we obtain
cu − cd ucd − dcu
α= ≥ 0, M= ≤ 0.
(u − d)S (u − d)R

Apparently, we are so fortunate to have 2 states of the world (for matching

the outcome) and two unknown α and M to be determined – equal number
of states and unknowns.

6
Query : What would happen in the above replicating procedure if the
discrete asset price process follows the trinomial random walk
model (3 states of the world in the next move)?

• Since M is always non-positive, the replicating portfolio involves buy-

ing the asset and borrowing cash in the corresponding proportions.

• The number of units of asset held is seen to be the ratio of the

difference of call values cu −cd to the difference of asset values uS −dS.
This is called the hedge ratio.

The call option can be replicated by a portfolio of basic securities:

risky asset and riskfree money market account. By invoking the no-
arbitrage principle, the call value is identical to the value of the repli-
cating portfolio.

7
Binomial option pricing formula

The current value of the call is given by the current value of the portfolio,
that is,
R−d c + u−R c
u−d u u−d d
c = αS + M =
R
pcu + (1 − p)cd R−d
= where p= .
R u−d

Note that the probability q, which is the subjective probability about

upward or downward movement of the asset price, does not appear in
the call value. The parameter p can be shown to be 0 < p < 1 since
u > R > d and so p can be interpreted as a probability.

8
Risk netural problem measure

From the relation

R−d u−R
puS + (1 − p)dS = uS + dS = RS,
u−d u−d
one can interpret the result as follows: the expected rate of returns on
the asset with p as the probability of upside move is just equal to the
riskless interest rate:
1
S = E ∗[S ∆t |S],
R
where E ∗ is expectation under this probability measure. We may view p
as the risk neutral probability .
" #
S ∆t
Since E ∗ S equals the current asset value S, we say that the dis-
R
counted asset value is a martingale under the risk neutral pricing measure.

9
Discounted expectation of the terminal payoff

The call price can be interpreted as the expectation of the payoff of the
call option at expiry under the risk neutral probability measure discounted
at the riskless interest rate.

The binomial call value formula can be expressed as

1 ∗ ∆t
c = E [c |S],
R
where c denotes the call value at the current time, and c∆t denotes the
random variable representing the call value one period later.

Besides applying the principle of replication of claims, the binomial option

pricing formula can also be derived using the riskless hedging principle or
via the concept of state prices.

10
Determination of the jump parameters

St+△t
• Under the risk neutral measure, ln becomes normally distributed
! St
σ2
with mean r − △t and variance σ 2△t, where r is the riskless
2
interest rate and σ 2 is the variance rate.
St+△t 2 σ 2 △t
• The mean and variance of are R and R (e −1), respectively,
St
where R = er△t.
• For the one-period binomial option model under the risk neutral mea-
St+△t
sure, the mean and variance of the asset price ratio are
S
pu + (1 − p)d and pu2 + (1 − p)d2 − [pu + (1 − p)d]2,
respectively.

11
 • By equating the mean and variance of the asset price ratio in both
continuous and discrete models, we obtain

pu + (1 − p)d = R
2
pu2 + (1 − p)d2 − R2 = R2(eσ △t − 1).
R−d
The first equation leads to p = , the usual risk neutral probability.
u−d
• A convenient choice of the third condition is the tree-symmetry con-
dition
1
u= ,
d
so that the lattice nodes associated with the binomial tree are sym-
2
metrical. Writing σe 2 = R2eσ △t, the solution is found to be
q
1 e2 + 1 +
σ e 2 + 1)2 − 4R2
(σ R−d
u= = , p= .
d 2R u−d

How to obtain a nice approximation to the above daunting expression?

12
 √
• By expanding u in Taylor series in powers of △t, we obtain
q σ2 4r2 + 4σ 2r + 3σ 4 3
u = 1 + σ △t + △t + △t 2 + O(△t2 ).
2 8σ
• Observe that the√
first three terms in the above Taylor series agree
with those of eσ △t up to O(△t) term.
• This suggests the judicious choice of the following set of parameter
values
√ √ R−d
u=e σ △t , d=e −σ △t, p= .
u−d
St+△t
• With this new set of parameters, the variance of the price ratio
St
in the continuous and discrete models agree up to O(△t).

13
Continuous limit of the binomial model

We consider the asymptotic limit △t → 0 of the binomial formula

c = [pc∆t ∆t −r△t.
u + (1 − p)cd ] e
In the continuous analog, the binomial formula can be written as

c(S, t − △t) = [pc(uS, t) + (1 − p)c(dS, t)] e−r△t .

Assuming sufficient continuity of c(S, t), we perform the Taylor expansion
of the binomial scheme at (S, t) as follows:

14
 −c(S, t − △t) + [pc(uS, t) + (1 − p)c(dS, t)]e−r△t
∂c 1 ∂ 2c 2 + · · · − (1 − e−r△t )c(S, t)
= (S, t)△t − (S, t)△t
∂t  2 ∂t2
−r△t ∂c
+ e [p(u − 1) + (1 − p)(d − 1)]S (S, t)
∂S
1 2 2 2 ∂ 2c
+ [p(u − 1) + (1 − p)(d − 1) ]S 2
(S, t)
2 ∂S )
1 3
∂ c
+ [p(u − 1)3 + (1 − p)(d − 1)3]S 3 3 (S, t) + · · · .
6 ∂S
By observing that
1 − e−r△t = r△t + O(△t2),
it can be shown that

e−r△t [p(u − 1) + (1 − p)(d − 1)] = r△t + O(△t2),

e−r△t [p(u − 1)2 + (1 − p)(d − 1)2] = σ 2△t + O(△t2),
e−r△t [p(u − 1)3 + (1 − p)(d − 1)3] = O(△t2).

15
Combining the results, we obtain
−r△t
−c(S,
" t − △t) + [pc(uS, t) + (1 − p)c(dS, t)] e #
∂c ∂c 2
σ 2 ∂ c2
= (S, t) + rS (S, t) + S (S, t) − rc(S, t) △t + O(△t 2).
∂t ∂S 2 ∂S 2

Since c(S, t) satisfies the binomial formula, so we obtain

∂c ∂c σ 2 2 ∂ 2c
0 = (S, t) + rS (S, t) + S 2
(S, t) − rc(S, t) + O(△t).
∂t ∂S 2 ∂S

In the limit ∆t → 0, the binomial call value c(S, t) satisfies the Black-
Scholes equation.

16
Multiperiod extension

Let cuu denote the call value at two periods beyond the current time with
two consecutive upward moves of the asset price and similar notational
interpretation for cud and cdd. The call values cu and cd are related to cuu ,
cud and cdd as follows:
pcuu + (1 − p)cud pcud + (1 − p)cdd
cu = and cd = .
R R

The call value at the current time which is two periods from expiry is
found to be
p2cuu + 2p(1 − p)cud + (1 − p)2cdd
c= 2
,
R
where the corresponding terminal payoff values are given by

cuu = max(u2S − X, 0), cud = max(udS − X, 0), cdd = max(d2 S − X, 0).

17
The coefficients p2, 2p(1 − p) and (1 − p)2 represent the respective risk
neutral probability of having two up jumps, one up jump and one down
jump, and two down jumps in two consecutive moves of the binomial
process.

Dynamics of asset price and call price in a two-period binomial model.

18
• With n binomial steps, the risk neutral probability of having j up jumps
n j n−j n n!
and n−j down jumps is given by Cj p (1−p) , where Cj =
j!(n − j)!
is the binomial coefficient.

• The corresponding terminal payoff when j up jumps and n − j down

jumps occur is seen to be max(uj dn−j S − X, 0).

• The call value obtained from the n-period binomial model is given by
n
X
Cjnpj (1 − p)n−j max(uj dn−j S − X, 0)
j=0
c= .
Rn

19
We define k to be the smallest non-negative integer such that uk dn−k S ≥
X
ln Sd n
X, that is, k ≥ u . It is seen that
ln d
(
0 when j<k
max(uj dn−j S − X, 0) = .
uj dn−j S − X when j≥k
The integer k gives the minimum number of upward moves required for
the asset price in the multiplicative binomial process in order that the call
expires in-the-money.

The call price formula is simplified as

n
X j dn−j n
X
u
c=S Cjnpj (1 − p)n−j − XR −n C npj (1 − p)n−j .
j
j=k
Rn j=k

20
Interpretation of the call price formula

• The last term in above equation can be interpreted as the expectation

value of the payment made by the holder at expiration discounted by
n
X
the factor R−n, and Cjnpj (1 − p)n−j is seen to be the probability
j=k
(under the risk neutral measure) that the call will expire in-the-money.

• The above probability is related to the complementary binomial dis-

tribution function defined by
n
X
Φ(n, k, p) = Cjnpj (1 − p)n−j .
j=k

Note that Φ(n, k, p) gives the probability for at least k successes in n trials
of a binomial experiment, where p is the probability of success in each
trial.

21
 up d(1 − p)
Further, if we write p′ = so that 1 − p′ = , then the call price
R R
formula for the n-period binomial model can be expressed as

c = SΦ(n, k, p′ ) − XR−n Φ(n, k, p).

• The first term gives the discounted expectation of the asset price at
expiration given that the call expires in-the-money.

• The second term gives the present value of the expected cost incurred
by exercising the call.

• In the replicating portfolio, we require long holding of Φ(n, k, p′) units

of the underlying asset and short holding of XR−nΦ(n, j, p) dollars of
the money market account.

22
The call price for the n-period binomial model can be expressed as the
discounted expectation of the terminal payoff under the risk neutral mea-
sure
1 1
c = n E ∗ [cT ] = n E ∗ [max(ST − X, 0)] , T = t + n∆t,
R R
where cT is the terminal payoff, max(ST − X, 0), of the call at expiration
1
time T and n is the discount factor over n periods. That is,
R
1 ∗
′
SΦ(n, k, p ) = E [ST 1{ST >X}]
Rn
Φ(n, k, p) = E ∗[1{ST >X}] = P ∗[ST > X].
The expectation operator E ∗ is taken under the risk neutral measure
rather than the true probability measure associated with the actual (phys-
ical) asset price process.

23
Dynamic programming procedure

American early exercise feature

• Without the early exercise privilege, risk neutral valuation leads to the
usual binomial formula
pV ∆t ∆t
u + (1 − p)V d
Vcont = .
R
• The following simple dynamic programming procedure is applied at
each binomial node

V = max(Vcont , h(S)),
where h(S) is the exercise payoff when the asset price assumes the
value S.

The optimally condition is applied at each binomial node.

24
American put option

• The intrinsic value of a vanilla put option is X − S n

j at the (n, j) node,
where X is the strike price. The dynamic programming procedure
applied at each node is given by
 
n+1 n+1
n  pP j+1 + (1 − p)P j 
P j = max  , X − Sn
j,
R

where n = N − 1, · · · , 0, and j = 0, 1, · · · , n. Here, N is the total

number of time steps in the binomial tree.

25
Example 1

Consider a 5-month American put option on a non-dividend-paying stock

when the stock price is $50, the strike price is $50, the risk-free interest
rate is 10% per annum, and the volatility is 40% per annum. That is,
S = 50, X = 50, r = 0.10, σ = 0.40, T = 0.4167. Suppose that we divide
the life of the option into five intervals of length 1 month (= 0.0833 year)
for the purpose of constructing a binomial tree. Then ∆t = 0.0833, we
have
√ √
u = eσ ∆t = 1.1224, d = e−σ ∆t = 0.8909, R = er∆t = 1.0084,
R−d
p= = 0.5073, 1 − p = 0.4927.
u−d

26
At each node:

Upper value = Underlying Asset Price

Lower value = Option Price

Shading indicates where option is exercised

Strike price = 50

Discount factor per step = 1/R = e−r∆t = 0.9917

Time step, ∆t = 0.0833 years, 30.42 days

Growth factor per step, R = 1.0084

Risk neutral probability of up move, p = 0.5073

Proportional up jump factor, u = 1.1224

Proportional down jump factor, d = 1/u = 0.8909

27
28
 • The stock price at the j th node (j = 0, 1, · · · , i) at time n∆t (n =
0, 1, · · · , 5) is calculated as S0uj dn−j . For example, the stock price at
node A (n = 4, j = 1) (i.e., the second node up at the end of the
fourth time step) is 50 × 1.1224 × 0.89093 = $39.69.

• The option prices at the final nodes are calculated as max(X − ST , 0).
For example, the option price at node G is 50.00 − 35.36 = 14.64.

Backward induction procedure

• First, we assume no exercise of the option at the nodes. This means

that the option price is calculated as the present value of the expected
option price one time step later. For example, at node E, the option
price is calculated as

(0.5073 × 0 + 0.4927 × 5.45)e−0.10×0.0833 = 2.66

whereas at node A it is calculated as

(0.5073 × 5.45 + 0.4927 × 14.64)e−0.10×0.0833 = 9.90.

29
Check to see if early exercise is preferable to waiting

• At node E, early exercise would give a value for the option of zero
because both the stock price and strike price are $50. Clearly it is
best to wait. The correct value for the option at node E, therefore,
is $2.66.

• At node A, it is a different story. If the option is exercised, it is worth

$50.00 − $39.69, or $10.31. This is more than $9.90. If node A is
reached, then the option should be exercised and the correct value
for the option at node A is $10.31.

• Option prices at earlier nodes are calculated in a similar way. Note

that it is not always best to exercise an option early when it is in the
money.

30
• Consider node B. If the option is exercised, it is worth $50.00−$39.69,
or $10.31. However, if it is held, it is worth

(0.5073 × 6.38 + 0.4927 × 14.64)e−0.10×0.0833 = 10.36.

The option should, therefore, not be exercised at this node, and the
correct option value at the node is $10.36.

• Working back through the tree, the value of the option at the initial
node is $4.49. This is our numerical estimate for the option’s current
value.

• In practice, a smaller value of ∆t, and many more nodes, would be

used. With 30, 50, 100, and 500 time steps we get values for the
option of 4.263, 4.272, 4.278, and 4.283.

31
Convergence of the price of the option

32
Callable American call

• The callable feature entitles the issuer to buy back the American
option at any time at a predetermined call price.
• Upon call, the holder can choose either to exercise the call or receive
the call price as cash.
• Let the call price be K. The dynamic programming procedure applied
at each node is modified as follows
  
n+1 n+1
  pC j+1 + (1 − p)C j 
Cjn = min max  , Sjn − X  ,
R


max(K, Sjn − X) .

33
  
n+1
pCn+1 + (1 − p)Cjn+1
• The first term max  , Sjn − X  represents the
R
optimal strategy of the holder, given no call of the option by the
issuer.
• Upon call by the issuer, the payoff is given by the second term
max(K, Sjn − X) since the holder can either receive cash amount K or
exercise the option.
• From the perspective of the issuer, he chooses to call or restrain
from calling so as to minimize the option value with reference to the
possible actions of the holder. The value of the callable call is given
by taking the minimum value of the above two terms.

34
Estimating delta and other Greek letters

• The delta (∆) of an option is the rate of change of its price with
respect to the underlying stock price. It can be calculated as
∆f
∆S
where ∆S is a small change in the stock price and ∆f is the corre-
sponding small change in the option price.

• At time ∆t, we have an estimate f11 for the option price when the
stock price is S0 u and an estimate f10 for the option price when the
stock price is S0d.

• When ∆S = S0u − S0d, ∆f = f11 − f10. Therefore an estimate of

delta at time ∆t is
f − f10
∆ = 11 .
S0 u − S0 d

35
Gamma calculations

To determine gamma (Γ), note that we have two estimates of ∆ at time

2∆t. When S = (S0u2 + S0)/2 (halfway between the second and third
node), delta is (f22 − f21)/(S0 u2 − S0 ); when S = (S0 + S0 d2)/2 (halfway
between the first and second node), delta is (f21 − f20)/(S0 − S0d2). The
difference between the two values of S is h, where

h = 0.5(S0u2 − S0 d2).

Gamma is the change in delta divided by h:

[(f22 − f21)/(S0 u2 − S0 )] − [(f21 − f20 )/(S0 − S0d2 )]
Γ= .
h

36
Theta calculations

Theta is the rate of change of the option price with time when all else is
kept constant. If the tree starts at time zero, an estimate of theta is
f21 − f00
Θ= .
2∆t
Note that f21 is the option value at two time steps from time zero and
with the same asset price.

Vega calculations

Vega can be calculated by making a small change, ∆σ, in the volatility

and constructing a new tree to obtain a new value of the option. (The
time step ∆t should be kept the same.) The estimate of vega is
f∗ − f
ν=
∆σ
where f and f ∗ are the estimates of the option price from the original
and the new tree, respectively.

37
Example 2

• Consider again Example 1. We have f1,0 = 6.96 and f1,1 = 2.16. An

estimate for delta is given by
2.16 − 6.96
= −0.41.
56.12 − 44.55
• An estimate of the gamma of the option can be obtained from the
values at nodes B, C, and F as
[(0.64 − 3.77)/(62.99 − 50.00)] − [(3.77 − 10.36)/(50.00 − 39.66)]
= 0.03.
11.65
• An estimate of the theta of the option can be obtained from the
values at nodes D and C as
3.77 − 4.49
= −4.3 per year
0.1667
or −0.012 per calendar day.

• These are only rough estimates. They become progressively better

as the number of time steps on the tree is increased.

38
Discrete dividend models

Consider the naive construction of the binomial tree. Let S be the asset
price at the current time which is n△t from expiry, and suppose a discrete
dividend of amount D is paid at time between one time step and two time
steps from the current time.

The nodes in the binomial tree at two time steps from the current time
would correspond to asset prices

u2S − D, S−D and d2 S − D,

since the asset price drops by the same amount as the dividend right after
the dividend payment.

39
• Extending one time step further, there will be six nodes

(u2S − D)u, (u2S − D)d, (S − D)u, (S − D)d, (d2S − D)u, (d2S − D)d
instead of four nodes as in the usual binomial tree without discrete
dividend.
• This is because (u2S − D)d 6= (S − D)u and (S − D)d 6= (d2 S − D)u,
so the interior nodes do not recombine.
• In general, suppose a discrete dividend is paid in the future between
kth and (k+1)th time step, then at the (k+m)th time step, the number
of nodes would be (m + 1)(k + 1) rather than k + 1 nodes as in the
usual reconnecting binomial tree.

40
Binomial tree with single discrete dividend.

41
• Splitting the asset price St into two parts: the risky component Set
that is stochastic and the remaining part that will be used to pay the
discrete dividend (assumed to be deterministic) in the future.
• Suppose the dividend date is t∗, then at the current time t, the risky
component Set is given by
( ∗
St − De−r(t −t) , t < t∗
Set =
St , t > t∗ .

• Let σ
e denote the volatility of S e and assume σ e to be constant rather
t
than the volatility of St itself to be constant.

42
• Assume that a discrete dividend D is paid at time t∗, which lies be-
tween the kth and (k + 1)th time step.
• At the tip of the binomial tree, the risky component Se is related to
the asset price S by
Se = S − De−kr∆t.

• The total value of asset price at the (n, j)th node, which corresponds
to n time steps from the tip and j upward jumps, is given by
e j dn−j + De−(k−n)r∆t 1
Su {n≤k},
n = 1, 2, · · · , N and j = 0, 1, · · · , n.

43
Construction of a reconnecting binomial tree with single discrete dividend
D. Here, N = 4 and k = 2, and let Se denote the risky component of the
asset value at the tip of the binomial tree. The asset value at nodes P, Q
and R are Se + De−2r∆t, Sue + De−r∆t and Sd,e respectively.

44
Example 3

Consider a 5-month American put option on a stock that is expected to

pay a single dividend of $2.06 during the life of the option. The initial
stock price is $52, the strike price is $50, the risk-free interest rate is 10%
per annum, the volatility is 40% per annum, and the ex-dividend date is
1
in 3 months.
2

Solution

We construct a tree to model Se (risky component of the asset price

process), the stock price less the present value of future dividends during
the life of the option. At time zero, the present value of the dividend is

2.06e−0.2917×0.1 = 2.00.

45
 • The initial value of Se is therefore 50.00.

e the figure
• Assuming that the 40% per annum volatility refers to S,
e
provides a binomial tree for S.

• Adding the present value of the dividend at each node leads to the
figure, which is a binomial model for S.

• The probabilities at each node are 0.5073 for an up movement and

0.4927 for a down movement. Working back through the tree in the
usual way gives the option price as 4.44.

Remark

Note that the exercise payoff is calculated using the actual asset price S,
not the risky component S.e

46
At each node:

Upper value = Underlying Asset Price

Lower value = Option Price

Shading indicates where option is exercised

Strike price = 50

Discount factor per step = 1/R = 0.9917

Time step, dt = 0.0833 years, 30.42 days

Growth factor per step, R = 1.0084

Risk neutral probability of up move, p = 0.5073

Proportional up jump factor, u = 1.1224

Proportional down jump factor, d = 1/u = 0.8909

47
48
Tree when stock pays a known dividend yield at one particular time. The
dividend amount is equal to δ times the prevailing asset price. In this
case, the interior nodes do recombine. Here, δ is the dividend yield.

49
Pricing of path dependent derivatives

• A path-dependent derivative is a derivative where the payoff depends

on the path followed by the price of the underlying asset, not just its
final value.

Two important properties:

1. The payoff from the derivative must depend on a single function, F ,

of the path followed by the underlying asset.

2. It must be possible to calculate the updated value of F at time τ + ∆t

from the known value of F at time τ and the updated value of the
underlying asset at time τ + ∆t.

50
American floating strike lookback put option on a non-dividend-paying
stock

• If exercised at time τ , this pays off the amount by which the maximum
stock price between time 0 and time τ exceeds the current stock price.
That is,
max Xt − Xτ .
t∈[0,τ ]

• We suppose that the initial stock price is $50, the stock price volatility
is 40% per annum, the risk-free interest rate is 10% per annum,
the total life of the option is three months, and that stock price
movements are represented by a three-step binomial tree. That is,
S0 = 50, σ = 0.4, r = 0.10, ∆t = 0.08333, u = 1.1224, d = 0.8909, R =
1.0084, and p = 0.5073.

51
Tree for valuing an American lookback option.

Rolling back through the tree gives the value of the American lookback
as $5.47.
52
• The top number at each node is the stock price. The next level
of numbers at each node shows the possible maximum stock prices
achievable on paths leading to the node. The final level of num-
bers shows the values of the derivative corresponding to each of the
possible maximum stock prices.

• The values of the derivatives at the final nodes of the tree are calcu-
lated as the maximum stock price minus the actual stock price.

• To illustrate the rollback procedures, suppose that we are at node A,

where the stock price is $50. The maximum stock price achieved thus
far is either 56.12 or 50 (depending on the path history of the asset
price movement). Consider first where it is equal to 50. If there is an
up movement, the maximum stock price becomes 56.12 and the value
of the derivative is zero. If there is a down movement, the maximum
stock price stays at 50 and the value of the derivative is 5.45.

53
• Assuming no early exercise, the value of the derivative at A when the
maximum achieved so far is 50 is,

(0 × 0.5073 + 5.45 × 0.4927)e−0.1×0.08333 = 2.66.

Clearly, it is not worth exercising at node A because the payoff from
doing so is zero.

• A similar calculation for the situation where the maximum value at

node A is 56.12 gives the value of the derivative at node A, without
early exercise, to be

(0 × 0.5073 + 11.57 × 0.4927)e−0.1×0.08333 = 5.65.

Early exercise gives a value of 6.12 and it is the optimal strategy.

• There may be multiple realized maximum asset values at each node.

The different possible values of the path dependent function at a
given node are linked to the corresponding path dependent function
at the nodes that are one time step earlier.

54
There are 2 possible realized maximum at node A, one is 50.00 while the
other is 56.12.

56.12 56.12
56.12 56.12
0.00 0.00

50.00 50.00
50.00 A 56.12 A
2.66 6.12

44.55 44.55
50.00 56.12
5.45 11.57

When the realized maximum at A When the realized maximum at

is 50.00, the realized maximum A is already 56.12, the realized
becomes 56.12 when the asset maximum remains at 56.12
price moves up while the realized independent of whether the
maximum remains at 50.00 when asset price moves up or down.
the asset price moves down.

55
1.2 Trinomial schemes

In a trinomial model, the asset price S is assumed to jump to either uS, mS

or dS after one time period △t, where u > m > d. We consider a trinomial
formula of option valuation of the form
p1V ∆t ∆t
u + p2V m + p3V d
∆t
V = , R = er△t.
R
This is deduced from the risk neutral valuation principle: the current
option value is the discounted expectation of the terminal option value
under the risk neutral pricing measure.

There are 6 unknowns: p1, p2, p3 , u, m and d. We take m = 1, u = 1/d.

We obtain 3 equations by

(i) equating mean, (ii) equating variance,

(iii) setting sum of probabilities = 1. We are left with one free parameter.

56
Discounted expectation approach

Under the assumption of the Geometric Brownian process followed by the

continuous asset price process, we write

ln St+△t = ln St + ζ,
!
σ2
where ζ is a normal random variable with mean r− △t and variance
2
σ 2△t. We approximate ζ by an approximate discrete random variable ζ a
with the following distribution

 v
 with probability p1
ζa = 0 with probability p2

 −v with probability p3
√
where v = λσ △t and λ ≥ 1. The corresponding values for u, m and d in
the trinomial scheme are: u = ev , m = 1 and d = e−v . This is because
St+∆t St+∆t
when ln assumes the value v, then = u assume the value ev .
St St

57
To find the probability values p1, p2 and p3 , the mean and variance of the
approximating discrete trinomial random walk variable ζ a are chosen to
be equal to those of ζ. These lead to
!
σ2
E[ζ a] = v(p1 − p3 ) = r − △t
2

var(ζ a) = v 2(p1 + p3 ) − v 2(p1 − p3)2 = σ 2 △t.

We see that v 2(p1 − p3)2 = O(∆t2). We may drop this term so that

v 2(p1 + p3) = σ 2 △t,

while still maintaining O(∆t) accuracy.

S S
By considering the approximation of log t+∆t instead of t+∆t , the al-
St St
gebraic equations for solving p1, p2 and p3 involve only linear functions of
∆t rather than exponential functions of ∆t.

58
Lastly, the probabilities must be summed to one so that

p1 + p2 + p3 = 1.
We then solve together to obtain
σ 2 √
1 (r − 2 ) △t
p1 = +
2λ2 2λσ
1
p2 = 1 − 2
λ
σ 2 √
1 (r − 2 ) △t
p3 = − ,
2λ2 2λσ
here λ is a free parameter.

• In order that p2 ≥ 0, we must choose λ ≥ 1.

√
• Numerical experiments indicate that the optimal choice of λ is 3 so
that p2 = 2/3.

59
• Note that p2 = 0 when λ = 1, which reduces to the Cox-Ross-
Rubinstein binomial scheme. This illustrates an effective mean of
deriving the binomial/trinomial parameters using the discrete approx-
imation of the logarithm of the price ratio at successive time steps.
 √
r− 2 σ2 ∆t
1
• When λ = 1, p1 = + . This would agree with the
2 2σ √
R−d
Taylor expansion of p = , u = 1/d = eσ ∆t up to O(∆t).
u−d

60
Multistate extension – Kamrad-Ritchken’s approach

• We assume the joint density of the prices of the two underlying assets
S1 and S2 to be bivariate lognormal.
• Let σi be the volatility of asset price Si, i = 1, 2 and ρ be the corre-
lation coefficient between the two lognormal diffusion processes.
△t
• Let Si and S i denote, respectively, the price of asset i at the current
time and one period △t later.
• Under the risk neutral measure, we have
△t
Si
ln = ζi , i = 1, 2,
Si
!
σ 2
where ζi is a normal random variable with mean r − i △t and vari-
2
ance σi2 △t.

61
The instantaneous correlation coefficient between ζ1 and ζ2 is ρ. The
joint bivariate normal process {ζ1, ζ2} is approximated by a pair of joint
discrete random variables {ζ1a, ζ2a} with the following distribution
{ 1 2}

ζ a1 ζ a2 probability
v1 v2 p1
v1 −v2 p2
−v1 −v2 p3
−v1 v2 p4
0 0 p5

√
where vi = λiσi △t, i = 1, 2.

The above form of the discrete distribution can be shown to be sufficient

to serve as the discrete approximation of the correlated diffusion processes
with drifts. It is redundant to include scenarios, like ζ1a = v1 and ζ2a = 0,
etc.

62
Equating the corresponding means gives
!
σ12
E[ζ1a] = v1(p1 + p2 − p3 − p4) = r− △t (i)
2
!
σ22
E[ζ2a] = v2(p1 − p2 − p3 + p4) = r− △t. (ii)
2

By equating the variances and covariance to O(△t) accuracy, we have

var(ζ1a) = v12(p1 + p2 + p3 + p4) = σ12△t (iii)

var(ζ2a) = v22(p1 + p2 + p3 + p4) = σ22△t (iv)

E[ζ1aζ2a] = v1v2(p1 − p2 + p3 − p4 ) = σ1σ2 ρ△t. (v)

In order that Eqs. (iii) and (iv) are consistent, we must set λ1 = λ2.

63
Writing λ = λ1 = λ2 , we have the following four independent equations
for the five probability values
σ12 √
(r − 2 ) △t
p1 + p2 − p3 − p4 =
λσ1

σ22 √
(r − 2 ) △t
p1 − p2 − p3 + p4 =
λσ2
1
p1 + p2 + p3 + p4 = 2
λ
ρ
p1 − p2 + p3 − p4 = 2 .
λ
Since the probabilities must be summed to one, this gives the remaining
condition as
p1 + p2 + p3 + p4 + p5 = 1.

64
The solution of the above linear algebraic system of equations gives
   
√ σ12 σ22
1 1 △t  r − 2 r− 2  ρ
p1 =  +  +  + 2
4 λ2 λ σ1 σ2 λ

   
√ σ12 σ22
1 1 △t  r − 2 r− 2  ρ
p2 =  +  −  − 
4 λ2 λ σ1 σ2 λ2

   
√ σ12 σ22
1 1 △t  r − 2 r− 2  ρ
p3 =  2+ − −  + 2
4 λ λ σ1 σ2 λ

   
√ σ12 σ22
1 1 △t  r − 2 r− 2  ρ
p4 =  2+ − +  − 2
4 λ λ σ1 σ2 λ

1
p5 = 1 − 2 , λ ≥ 1 is a free parameter.
λ

65
Two-state trinomial model

• For convenience, we write ui = evi , di = e−vi , i = 1, 2.

• Let V ∆t
u1 u2 denote the option price at one time period later with asset
prices u1S1 and u2S2 , and similar meaning for V ∆t ∆t ∆t
u d , V d u and V d d .
1 2 1 2 1 2

• We let V ∆t
0,0 denote the option price one period later with no jumps
in asset prices.
• The corresponding 5-point formula for the two-state trinomial model
based on the risk neutral valuation approach can be expressed as
△t △t △t △t
V = (p1Vu△t
1 u2
+ p2 Vu d + p3Vd d + p4Vd u + p5V 0,0)/R.
1 2 1 2 1 2

• When λ = 1, we have p5 = 0 and the above 5-point formula reduces

to the 4-point formula.

66
1.3 Forward shooting grid methods (strongly path dependent op-
tions)

• For path dependent options, the option value also depends on the
path function Ft = F (S, t) defined specifically for the given nature
of path dependence, say, the minimum asset price realized along a
specific asset price path.
• Since option value depends also on Ft, we find the value of the path
dependent option at each node in the lattice tree for all alternative
values of Ft that can occur.
• The approach of appending an auxiliary state vector at each node
in the lattice tree to model the correlated evolution of Ft with St is
commonly called the forward shooting grid (F SG) method.

67
• Consider a trinomial tree whose probabilities of upward, zero and
downward jump of the asset price are denoted by pu, p0 and pd, re-
spectively.
n denote the numerical option value of the exotic path depen-
• Let Vj,k
dent option at the nth-time level (n time steps from the tip of the
tree). Also, j denotes the j upward jumps from the initial asset value
and k denotes the numbering index for the various possible values of
the augmented state variable Ft at the (n, j)th node.
• Let G denote the function that describes the correlated evolution of
Ft with St over the time interval ∆t, that is,

Ft+∆t = G(t, Ft, St+∆t ).

68
• Let g(k, n, j) denote the grid function which is considered as the dis-
crete analog of the evolution function G.

• The trinomial version of the FSG scheme can be represented as follows

h i
n+1 n+1 n+1
Vj,k = pu Vj+1,g(k,n,j+1) + p0Vj,g(k,n,j) + pdVj−1,g(k,n,j−1) e−r∆t,
n

where e−r∆t is the discount factor over time interval ∆t.

• To price a specific path dependent option, the design of the FSG

algorithm requires the specification of the grid function g(k, n, j).

For notational convenience, if the grid function has no dependence

on t, we simply write it as g(k, j).

69
Cumulative Parisian feature

• Let M denote the prespecified number of cumulative breaching occur-

rences that is required to activate knock-out, and let k be the integer
variable that counts the cumulative number of breaching occurrences
so far.

• Let B denote the down barrier associated with the knock-out feature.

• Let xj denote the value of x = ln S that corresponds to j upward

moves in the trinomial tree. That is, xj = ln S0 + j∆x, where S0 is
the initial asset price and ∆x is the stepwidth of the state variable x.

• When n∆t happens to be a monitoring instant, the index k increases

its value by 1 if the asset price S falls on or below the barrier B, that
is, xj ≤ ln B.

70
To incorporate the cumulative Parisian feature, the appropriate choice of
the grid function gcum(k, j) is defined by

gcum(k, j) = k + 1{xj ≤ln B}.

The backward induction procedure in the trinomial tree calculations is
exemplified by

 [p V n + p V n +p Vn ]e −r∆t

 u j+1,k 0 j,k d j−1,k


n−1 if n∆t is not a monitoring instant
Vj,k = n n n −r∆t .


 [p V
u j+1,gcum (k,j+1) + p V
0 j,gcum (k,j) + p V
d j−1,gcum (k,j−1) ]e


if n∆t is a monitoring instant

The number of breaching occurrences k is updated to gcum(k, j + 1) when

n
the updated asset price at the nth time level is Sj+1 [up move from Sjn−1
at the (n − 1)th time level].

71
Schematic diagram that illustrates the construction of the grid function
gcum(k, j) that models the cumulative Parisian feature. The down barrier
ln B is placed mid-way between two horizontal rows of trinomial nodes.
Here, the nth-time level is a monitoring instant.

72
Remarks

1. The pricing of options with the continuously monitored cumulative

Parisian feature is obtained by setting all time steps to be monitoring
instants.

2. The computational time required for pricing an option with the cu-
mulative Parisian feature requiring M breaching occurrences to knock
out is about M times that of an one-touch knock-out barrier option.

3. The consecutive Parisian feature counts the number of consecutive

breaching occurrences that the asset price stays in the knock-out
region. The count is reset to zero once the asset price moves out
from the knock-out region. Assuming B to be the down barrier, the
appropriate grid function gcon (k, j) in the FSG algorithm is given by

gcon (k, j) = (k + 1)1{xj ≤ln B} .

73
Call options with strike reset feature

• Consider a call option with the strike reset feature where the option’s
strike price is reset to the prevailing asset price on a preset reset date
if the option is out-of-money on that date.

• Let ti, i = 1, 2, · · · , M , denote the reset dates and Xi denote the strike
price specified on ti based on the above reset rule.

• Write X0 as the strike price set at initiation, then Xi is given by

Xi = min(Xi−1 , Sti ), i = 1, 2, · · · , M,
where Sti is the prevailing asset price at reset date ti.

• Why does it become superfluous to set

Xi = min(Xi−1 , Sti , X0 ), i = 1, 2, · · · , M ?
Since X1 = min(X0, St1 , the information of the initial strike price X0
has been embedded in the strike reset procedure.

74
• The strike price at expiry of this call option is not fixed since its value
depends on the realization of the asset price at the reset dates.

• When we apply the backward induction procedure in the trinomial

calculations, we encounter the difficulty in defining the terminal payoff
since the strike price is not yet known.

• These difficulties can be resolved easily using the FSG approach by

tracking the evolution of the asset price and the strike reset through
an appropriate choice of the grid function.

75
• Suppose the original strike price X0 corresponds to the index k0, this
would mean X0 = S0 uk0 . For convenience, we may choose the pro-
portional jump parameter u such that k0 is an integer. In terms of
these indexes, the grid function that models the correlated evolution
between the reset strike price and asset price is given by

greset(k, j) = min(k, j),

where k denotes the index that corresponds to the strike price reset in
the last reset date and j is the index that corresponds to the prevailing
asset price at the reset date.

• Since the strike price is reset only on a reset date, we perform the usual
trinomial calculations for those time levels that do not correspond
to a reset date while the augmented state vector of strike prices are
adjusted according to the grid function greset(k, j) for those time levels
that correspond to a reset date.

76
• The FGS algorithm for pricing the reset call option is given by

 h i
 n+1 n+1 n+1 −r∆t



pu Vj+1,k + p0 Vj,k + pd Vj−1,k e







 if (n + 1)∆t 6= ti for some i

n
Vj,k = h i .

 n+1 n+1 n+1

 p V + p V + p V e −r∆t,

 u j+1,greset(k,j+1) 0 j,greset(k,j) d j−1,greset(k,j−1)






 if (n + 1)∆t = ti for some i

• The payoff values along the terminal nodes at the N th time level in
the trinomial tree are given by
N = max(S uj − S uk , 0),
Vj,k j = −N, −N + 1, · · · , N,
0 0

and k assumes values that lie between k0 and the index corresponding
to the lowest asset price on the last reset date.

77
Floating strike arithmetic averaging call

• To price an Asian option, we find the option value at each node for
all possible values of the path function F (S, t) that can occur at that
node.

• Unfortunately, the number of possible values for the averaging value F

at a binomial node for arithmetic averaging option grows exponentially
at 2n, where n is the number of time steps from the tip of the binomial
tree. (Why 2n? Since there are 2n possible realized asset paths after
n time steps and each path gives a unique arithmetic averaging value.)

• Therefore, the binomial schemes that place no constraint on the num-

ber of possible F values at a node become computationally infeasible.

78
Illustration

Consider the following tree

62.99 62.99
56.12 56.12

50.00 50.00 50.00 50.00

44.55 44.55
39.69 39.69

There are 4 = 22 possible arithmetic averaging values after 2 time steps,

namely,
50.00 + 56.12 + 62.99 50.00 + 56.12 + 50.00
Auu = , Aud = ,
3 3
50.00 + 44.55 + 50.00 50.00 + 44.55 + 39.69
Adu = , Add = .
3 3

79
Note that these arithmetic averaging values do not coincide with the
stock prices at the nodes at the 2nd time level. Extending to a 3-step bi-
nomial tree, there are 8 = 23 possible arithmetic averaging values, namely,
Auuu , Auud, Audu, · · · , Addd.

Geometric averaging values

• Two-step binomial tree

q q
Guu = 3
(S0 )(S0 u)(S0u2) = S0 u, Gdd = 3
S0 (S0u−1)(S0 u−2) = S0 u−1,
q
Gud = 3
(S0 )(S0 u)(S0 ) = S0u1/3,
q
3
Gdu = (S0 )(S0 u−1)(S0 ) = S0u−1/3.
These 3 geometric averaging values coincide with the stock prices at
the nodes at the 2nd time level.

80
• Three-step binomial tree
q
(S0 )(S0 u)(S0u2)(S0 u3) = S0 u1.5,
4
Guuu =
q
(S0 )(S0 u−1)(S0 u−2)(S0 u−3) = S0 u−1.5,
4
Gddd =
q
4
Guud = (S0 )(S0 u)(S0 u2)(S0 u) = S0u,
Gudu = S0 u0.5, Gduu = S0 u0.25,
q
4
Gudd = (S0)(S0 u)(S0 )(S0 u−1) = S0 ,
Gdud = S0u−0.5, Gddu = S0u−1.
There are 8 possible geometric averaging values after 3 time steps.

Question How many possible goemetric averaging values after n time

steps?

81
• A possible remedy is to restrict the possible values for F to a certain
set of predetermined values. The option value V (S, F, t) for other
values of F is obtained from the known values of V at predetermined
F values by interpolation between the nodal values.

• The methods of interpolation include the nearest node interpolation,

linear (between 2 neighboring nodes) and quadratic interpolation (be-
tween 3 neighboring nodes).

• How to cope with the exponentially large number of possible values

assumed by arithmetic averaging of the realized asset price path? We
limit the number of averaging values to some multiple of the number
of values assumed by the asset price (here, the multiple is 1/ρ).

82
For a given time step ∆t, we fix the stepwidths to be
√
∆W = σ ∆t and ∆Y = ρ∆W, ρ < 1,
and define the possible values for St and At at the nth time step by

Sjn = S0 ej∆W and An

k = S0 e
k∆Y
,
where j and k are integers, and S0 is the asset price at the tip of the
binomial tree. We take 1/ρ to be an integer. The larger integer value
chosen for 1/ρ, the finer the quantification of the arithmetic averaging
asset value.

83
 !
1
Quantification of arithmetic averaging asset value Here, = 3 is taken.
ρ

after 2 time steps after 2 time steps

S 0 e 2 'W A00 e 6 'Y

S 0 e 'W A00 e3'Y

S0 S0 A0 A00
0

S0e  'W A00 e 3'Y

S 0 e 2 'W A00 e 6 'Y

stock price averaging price

84
 • The continuous version of the arithmetic averaging state variable is
defined by
Z
1 t
At = Su du.
t 0
The terminal payoff of the floating strike Asian call option is given by
max(ST − AT , 0), where AT is the arithmetic average of St over the
time period [0, T ].

Consider
d(tAt) = St dt,
we approximate d(tAt) at time t + ∆t by [(t + ∆t) + ∆t]At+At − (t + ∆t)At ,
so that
(t + ∆t)At + ∆t St+∆t
At+∆t = ≡ G(t, At , St+∆t ).
t + 2∆t
This is the updating rule of At+∆t dt at the new time level t + ∆t based
on the old value At at the previous time level t and updated asset value
St+∆t at the new time level t + ∆t.

85
Consider the binomial procedure at the (n, j)th node, suppose we have
n+1
an upward move in asset price from Sjn to Sj+1 and let An+1 be the
k+(j)
corresponding updated value of At changing from An k when the asset price
n+1
moves up from Sjn to Sj+1 . Setting A0
0 = S0, the equivalence of the above
equation is given by
n+1
(n + 1)An
k + Sj+1
An+1
+ = .
k (j) n+2

n+1
• For a downward move in asset price from Sjn to Sj−1 , An
k changes to
An+1
k−(j)
where
n+1
(n + 1)An
k + Sj−1
An+1
k−(j)
= .
n+2
′
Note that An+1
±
k (j)
in general do not coincide with A n+1
k ′ = S0 e k ∆Y , for

some integer k′.

86
• We define the integers kf±loor such that An+1
± are the largest possible
kfloor
An+1
k ′ values less than or equal to A n+1
k±(j)
. Accordingly, we compute
the indexes k±(j) by
(n+1)ek∆Y +e(j±1)∆W
ln n+2
k±(j) = . (1)
∆Y

• We then set kf+loor = floor(k+(j)) and kf−loor = floor(k−(j)), where

floor(x) denotes the largest integer less than or equal to x. Equation
(1) corresponds to the evalution of An n
k to Ak±(j) based on updated
n+1
Sj±1 with reference to k and k±(j).

87
• What would be the possible range of k at the nth time step? We ob-
serve that the arithmetic averaging state variable At must lie between
the maximum asset value Snn and the minimum asset value S−n n , so k

must lie between − nρ ≤ k ≤ n . Unless ρ assumes a very small value,

ρ
the number of predetermined values for At is in general manageable.

• Consider An ℓ , where ℓ is in general a real number. We write ℓf loor =

floor(ℓ) and let ℓceil = ℓf loor + 1, then An
ℓ lies between A n
ℓfloor and A n .
ℓceil
Though the number of possible values of ℓ grows exponentially with
the number of time steps in the binomial tree, both ℓf loor and ℓceil at
th n n
the n time level assume an integer value lying between − and .
ρ ρ

88
Linear interpolation

• Let cn th
j,ℓ denote the Asian call value at the (n, j) node with the aver-
aging state variable assuming the value An ℓ , and similar notations for
cn
j,ℓ and cn n
j,ℓ . Note that cj,ℓ is not defined if ℓ is not an integer.
floor ceil

• For a non-integer value ℓ, cn j,ℓ is approximated through interpolation

using the call values at the neighboring nodes. We approximate cn j,ℓ in
terms of cn
j,ℓ and cn
j,ℓ by the following linear interpolation formula
floor ceil

cn
j,ℓ = ǫℓ cn
j,ℓceil + (1 − ǫℓ )cn
j,ℓfloor ,
where
ln An
ℓ − ln A n
ℓfloor
ǫℓ = .
∆Y
Here, ǫℓ is the fractional step between ℓf loor and ℓceil , where
ǫℓ∆Y
An
ℓ = A n
ℓfloor e .

89
 H"
Anfloor (" ) A"n n
Aceil (")
x x x

• Here, ℓ is a real number lying between two consecutive integers floor(ℓ)

and ceil(ℓ), where ceil(ℓ) = floor(ℓ) + 1.

• Numerical option values are available only at An

f loor(ℓ)
and A n
ceil(ℓ)
,
where the index k in Ank assumes an integer value [like floor(ℓ) or
ceil(ℓ)].

• For ℓ to be a non-integer, we approximate cn

j,ℓ by linear interpolation
between cn
j,f loor(ℓ)
and cn
j,ceil(ℓ)
.

90
• By applying the above linear interpolation formula [taking ℓ to be
k+(j) and k−(j) successively], the FSG algorithm with linear interpo-
lation for pricing the floating strike arithmetic averaging call option is
given by
 
n+1
cn
j,k = e −r∆t
pc + (1 − p)cn+1
j−1,k− (j)
j+1,k+ (j)
( " #
= e−r∆t p ǫk+(j) cn+1 + + (1 − ǫk+ (j))cn+1 +
j+1,kceil j+1,kfloor
" #)
+ (1 − p) ǫk− (j) cn+1 − + (1 − ǫk− (j))cn+1 − (2)
j−1,kceil j−1,kfloor
n
n = N − 1, · · · , 0, j = −n, −n + 2, · · · , n, k is an integer between − and
ρ
n ±
, k (j) are given by Eq. (i) while
ρ

ln An+1
±
k (j)
− ln A n+1
±
kfloor
ǫk± (j) = . (3)
∆Y

91
The final condition is

cN
j,k = max(Sj
N
− A N
k , 0)
= max(S0 ej∆W − S0ek∆Y , 0), j = −N, −N + 2, · · · , N,
N N
and k is an integer between − and .
ρ ρ

Since the range of averaging values is narrower than that of the asset
N
prices, the range of k should be narrower than the range between −
ρ
N
and .
ρ

92
• At each terminal node (N, j), j = −N, −N + 2, · · · , N , we compute all
possible payoff values of the Asian call option with varying values of
k.

• To proceed with the backward induction procedure, at a typical (n, j)th

node, we find all possible call values with varying integer value k lying
n n
between − and using Eq. (2).
ρ ρ

• For a given integer value k, we compute k±(j) and ǫk± (j) using Eq.
(1) and Eq. (3), respectively.

93
 ­° Aceil
n 1
( k  ( j ))
® n 1
S nj S 0u j S 0 e j'W x °̄C j ,ceil ( k  ( j ))
Akn S 0 e k'Y S 0 e kU'W 1  H k  ( j ) ­ S n 1
° nj11
® Ak  ( j )
°c n 1
¯ j ,k  ( j )
x­ n 1
° floor ( k  ( j ))
A
® n 1
­S n °̄C j , floor ( k  ( j ))
j
° n
®A k
°c n
¯ j ,k ­° Aceil
n 1
( k  ( j ))
® n 1
°̄C j ,ceil ( k  ( j ))
x

­ S nj11
° n 1
® Ak  ( j )
Hk °c n 1

( j) ¯ j ,k  ( j )
­° Anfloor
1
x
( k  ( j ))
® n 1
°̄C j , floor ( k  ( j ))
94
In summary,

An −→ A n when S n −→ S n+1
k +
k (j) j j+1
An −→ A n when S n −→ S n+1
k −
k (j) j j−1

Note that k is an integer while k+(j) and k−(j) are in general non-integers.

cn +
j,k (j)
= ǫ + cn
+
k (j) j,f loor(k (j)) + (1 − ǫ )cn
k (j) j,ceil(k+(j)) ,
+

where
ln An+ − ln An
k (j) f loor(k+ (j))
ǫk+(j) = .
∆Y
Using the discounted expectation approach, we have
 
n = pC n+1
Cj,k n+1
+ (1 − p)Cj−1,k −r∆t.
j+1,k+(j) − (j) e

95
Alpha Quantile Options

The α-quantile option takes the barrier level to be a stochastic state

variable that defines the terminal payoff.

• For a given percentile α, 0 ≤ α < 1, the α-percentile of {St }t∈[0,T ] is

defined as
( Z )
1 T
Binf (T ; α) = inf B : 1{St≤B} dt > α .
T 0
Binf (T ; α) is the barrier level such that the asset price St is below
Binf (T ; α) over exactly α-portion of the monitoring period. When
α = 0.5, Binf (T ; 0.5) is the median Smedian of the asset price process
over the monitoring period.

96
 asset price

Binf(T; 1)

Smedian =
Binf(T; 0.5)

Binf(T; 0)

time
T

97
 Z
1 T
• 1{St≤B} dt is an increasing function of B.
T 0

• The asset price is below Smedian exactly half of the time period [0, T ].

• Binf (T ; 1−) is the realized maximum asset price over [0, T ] since the
asset price is below this barrier level 100% of the time period.
Z
1 T
−
• For any barrier level B higher than Binf (T ; 1 ), we have 1{St≤B} dt = 1.
T 0
−
In fact, Binf (T ; 1 ) is the infimum among all these barrier levels.

• Binf (T ; 0) is the realized minimum asset price over [0, T ] since it is

Z
1 T
the infimum among all barrier levels B such that 1{St≤B} dt > 0.
Z T 0
1 T
Note that for B < Binf (T ; 0), we have 1{St≤B} dt = 0.
T 0

98
For a European α-quantile call option, the terminal payoff is given by

Vα(S, T ) = max(Binf (T ; α) − X, 0),

where X is the strike price.

• In the discrete trinomial tree model with N time steps, we write SjN as
the discrete terminal asset price at maturity, j = −N, −N + 1, · · · , N .
The possible values taken by the stochastic variable Binf are limited to
Sj , j = −N, · · · , N −1, N ; Sj = S0 uj , where u is the up jump parameter.

• The numerical approximate value of the continuously monitored Eu-

ropean α-quantile call option at time 0 is given by
N
X
Vα (S, 0) = e−rT P [Binf (T ; α) = Sj ] max(Sj − X, 0), Sj = S0uj .
j=−N

99
How to compute Binf (T ; α)?

bin (d, B) denote the state price of an option that pays $1 at

• Let Vcum
maturity T if the cumulative time staying above the down-barrier B
is more than d of the total length of the time period, 0 < d < 1;
otherwise, the payoff is zero.

• In the discrete world of the trinomial tree model, we have

e−rT P [Binf (T ; α) > Sj ] = Vcum

bin [(1 − α)T, S ]
j

so that
n h io
−rT −rT
e P [Binf (T ; α) = Sj ] = e P [Binf (T ; α) > Sj−1] − P Binf (T ; α) > Sj
bin [(1 − α)T, S
= Vcum bin
j−1] − Vcum [(1 − α)T, Sj ].

100