Notes in Financial Engineering

Product of Geometric Brownian Motion Processes
Let dY /Y dZ/Z = a dt + b dWY , = f dt + g dWZ .
Consider the Ito process U Y Z . Apply Itos lemma (Theorem 20 on p. 510):

dU = = Z dY + Y dZ + dY dZ ZY (a dt + b dWY ) + Y Z (f dt + g dWZ ) +Y Z (a dt + b dWY )(f dt + g dWZ ) = U (a + f + bg) dt + U b dWY + U g dWZ .
c 2013 Prof. Yuh-Dauh Lyuu, National Taiwan University
Page 514
Product of Geometric Brownian Motion Processes (continued)

The product of two (or more) correlated geometric Brownian motion processes thus remains geometric Brownian motion. Note that Y Z U = exp a b /2 dt + b dWY , [( ) ] 2 = exp f g /2 dt + g dWZ , [( ( 2 ) ) ] 2 = exp a + f b + g /2 dt + b dWY + g dWZ .
2
[(
Page 515
Product of Geometric Brownian Motion Processes (concluded)

ln U is Brownian motion with a mean equal to the sum of the means of ln Y and ln Z . This holds even if Y and Z are correlated. Finally, ln Y and ln Z have correlation .
Page 516
Quotients of Geometric Brownian Motion Processes

Suppose Y and Z are drawn from p. 514. Let U Y /Z . We now show thata dU = (a f + g 2 bg) dt + b dWY g dWZ . U
(57)
Keep in mind that dWY and dWZ have correlation .

a Exercise
14.3.6 of the textbook is erroneous.
Page 517
Quotients of Geometric Brownian Motion Processes (concluded)

The multidimensional Itos lemma (Theorem 20 on p. 510) can be employed to show that
dU = = (1/Z ) dY (Y /Z 2 ) dZ (1/Z 2 ) dY dZ + (Y /Z 3 ) (dZ )2 (1/Z )(aY dt + bY dWY ) (Y /Z 2 )(f Z dt + gZ dWZ ) (1/Z 2 )(bgY Z dt) + (Y /Z 3 )(g 2 Z 2 dt) = U (a dt + b dWY ) U (f dt + g dWZ ) U (bg dt) + U (g 2 dt) = U (a f + g 2 bg) dt + U b dWY U g dWZ .
Page 518
Forward Price
Suppose S follows dS = dt + dW. S Consider F (S, t) Sey(T t) . Observe that F S 2F S 2 F t = ey(T t) , = 0, = ySey(T t) .
Page 519
Forward Prices (concluded)

Then dF = ey(T t) dS ySey(T t) dt = Sey(T t) ( dt + dW ) ySey(T t) dt = F ( y ) dt + F dW by Eq. (55) on p. 509. Thus F follows dF = ( y ) dt + dW. F This result has applications in forward and futures contracts.a
a It
is also consistent with p. 464.
Page 520
Ornstein-Uhlenbeck Process
The Ornstein-Uhlenbeck process: dX = X dt + dW, where , 0. It is known that
E [ X (t) ] Var[ X (t) ] Cov[ X (s), X (t) ] = = = e
(tt0 )
E [ x0 ] ,
) 2 ( 2(tt0 ) 2(tt0 ) 1e +e Var[ x0 ], 2 ] 2 (ts) [ 2(st0 ) e 1e 2 +e

(t+s2t0 )
Var[ x0 ],
for t0 s t and X (t0 ) = x0 .
Page 521
Ornstein-Uhlenbeck Process (continued)

X (t) is normally distributed if x0 is a constant or normally distributed. X is said to be a normal process. E [ x0 ] = x0 and Var[ x0 ] = 0 if x0 is a constant. The Ornstein-Uhlenbeck process has the following mean reversion property. When X > 0, X is pulled toward zero. When X < 0, it is pulled toward zero again.
Page 522
Ornstein-Uhlenbeck Process (continued)

A generalized version: dX = ( X ) dt + dW, where , 0. Given X (t0 ) = x0 , a constant, it is known that E [ X (t) ] Var[ X (t) ] for t0 t. = = + (x0 ) e(tt0 ) , ] 2 [ 2(tt0 ) 1e , 2 (58)
Page 523
Ornstein-Uhlenbeck Process (concluded)

The mean and standard deviation are roughly and / 2 , respectively. For large t, the probability of X < 0 is extremely unlikely in any nite time interval when > 0 is large relative to / 2 . The process is mean-reverting. X tends to move toward . Useful for modeling term structure, stock price volatility, and stock price return.
Page 524
Square-Root Process
Suppose X is an Ornstein-Uhlenbeck process. Itos lemma says V X 2 has the dierential, dV = 2X dX + (dX )2 = 2 V ( V dt + dW ) + 2 dt ( ) 2 = 2V + dt + 2 V dW,
a square-root process.
Page 525
Square-Root Process (continued)

In general, the square-root process has the stochastic dierential equation, dX = ( X ) dt + X dW, where , 0 and X (0) is a nonnegative constant. Like the Ornstein-Uhlenbeck process, it possesses mean reversion: X tends to move toward , but the volatility is proportional to X instead of a constant.
Page 526
Square-Root Process (continued)

When X hits zero and 0, the probability is one that it will not move below zero. Zero is a reecting boundary. Hence, the square-root process is a good candidate for modeling interest rates.a The Ornstein-Uhlenbeck process, in contrast, allows negative interest rates. The two processes are related (see p. 525).
a Cox,
Ingersoll, and Ross (1985).
Page 527
Square-Root Process (concluded)

The random variable 2cX (t) follows the noncentral chi-square distribution,a ( ) 4 t , 2 cX (0) e , 2 where c (2/ 2 )(1 et )1 . Given X (0) = x0 , a constant, ) ( t t E [ X ( t) ] = x 0 e , + 1e ) ) 2 ( 2 ( t 2t t 2 e + 1e , Var[ X (t) ] = x0 e 2 for t 0.
a William
Feller (19061970) in 1951.
Page 528
Modeling Stock Prices

The most popular stochastic model for stock prices has been the geometric Brownian motion, dS = dt + dW. S The continuously compounded rate of return X ln S follows dX = ( 2 /2) dt + dW by Itos lemma.a
a See
also Eq. (56) on p. 513. Also consistent with Lemma 9 (p. 259).
Page 529
Modeling Stock Prices (continued)

The more general deterministic volatility model posits dS = (rt qt ) dt + (S, t) dW, S where (S, t) is called the local volatility function.a The trees for the deterministic volatility model are called implied trees.b Their construction requires option prices at all strike prices and maturities.
and Kani (1994). b Derman and Kani (1994); Dupire (1994); Rubinstein (1994).
a Derman
Page 530

How to construct an implied tree with eciency, valid probabilities and stability remains open for a long time.a Reasons may include: noise and nonsynchrony in data, arbitrage opportunities in the smoothed and interpolated/extrapolated implied volatility surface, wrong model, etc. Numerically, the inversion problem is an ill-posed problem. It is partially solved recently.b
and Kani (1994); Derman, Kani, and Chriss (1996); Coleman, Kim, Li, and Verma (2000); Ayache, Henrotte, Nassar, and Wang (2004); Kamp (2009). b February 12, 2013.
a Derman
Page 531

Theoretically,a (X, T ) = 2
2 C T C + (rT qT )X X + qT C 2C X 2 X 2
(59)
where C is the call price at time t with strike price X and maturity date T . In practice, like implied trees and for similar reasons, (K, T )2 may have spikes, vary wildly, or even be negative. Worse, the appearance of 2 C/X 2 in the denominator results in numerical instability.
a Dupire
(1994); Andersen and Brotherton-Ratclie (1998).
Page 532
Modeling Stock Prices (concluded)

The relation between the implied volatility surface (X, T ) and the local volatility surface (X, T ) isa
[ ] + 2 T + (rT qT )X X [ ], (X, T )2 = ( ) ( )2 Xy 2 X 2 1 X + X X 4 + X X 2 X
2
where T t and y ln(X/St ) +
T
t
(qs rs ) ds.
Although this version may be more stable than Eq. (59) on p. 532, it is expected to suer from similar problems.
(1996); Andersen and Brotherton-Ratclie (1998); Gatheral (2003); Wilmott (2006); Kamp (2009).
a Andreasen
Page 533
Continuous-Time Derivatives Pricing
Page 534
I have hardly met a mathematician who was capable of reasoning. Plato (428 B.C.347 B.C.) Fischer [Black] is the only real genius Ive ever met in nance. Other people, like Robert Merton or Stephen Ross, are just very smart and quick, but they think like me. Fischer came from someplace else entirely. John C. Cox, quoted in Mehrling (2005)
Page 535
Toward the Black-Scholes Dierential Equation

The price of any derivative on a non-dividend-paying stock must satisfy a partial dierential equation (PDE). The key step is recognizing that the same random process drives both securities. As their prices are perfectly correlated, we gure out the amount of stock such that the gain from it osets exactly the loss from the derivative. The removal of uncertainty forces the portfolios return to be the riskless rate. PDEs allow many numerical methods to be applicable.
Page 536
Assumptionsa
The stock price follows dS = S dt + S dW . There are no dividends. Trading is continuous, and short selling is allowed. There are no transactions costs or taxes. All securities are innitely divisible. The term structure of riskless rates is at at r. There is unlimited riskless borrowing and lending. t is the current time, T is the expiration time, and T t.
and Taleb (2005) summarizes criticisms on these assumptions and the replication argument.
a Derman
Page 537
Black-Scholes Dierential Equation

Let C be the price of a derivative on S . From Itos lemma (p. 506), ( ) 2 C C C 1 C dC = S + + 2 S 2 dt + S dW. 2 S t 2 S S The same W drives both C and S . Short one derivative and long C/S shares of stock (call it ). By construction, = C + S (C/S ).
Page 538
Black-Scholes Dierential Equation (continued)

The change in the value of the portfolio at time dt isa C d = dC + dS. S Substitute the formulas for dC and dS into the partial dierential equation to yield ) ( 2 1 C C 2 S 2 d = dt. 2 t 2 S As this equation does not involve dW , the portfolio is riskless during dt time: d = r dt.
a Mathematically
speaking, it is not quite right (Bergman, 1982).
Page 539
Black-Scholes Dierential Equation (concluded)

So ( C 1 C + 2 S 2 t 2 S 2
2
( ) C dt = r C S dt. S
Equate the terms to nally obtain C C 1 2 2 2C + rS + S = rC. 2 t S 2 S When there is a dividend yield q , C 1 2 2 2C C + (r q ) S + S = rC. t S 2 S 2
Page 540
Rephrase
The Black-Scholes dierential equation can be expressed in terms of sensitivity numbers, + rS + 1 2 2 S = rC. 2 (60)
Identity (60) leads to an alternative way of computing numerically from and . When a portfolio is delta-neutral, 1 2 2 + S = rC. 2 A denite relation thus exists between and .
Page 541
[Black] got the equation [in 1969] but then was unable to solve it. Had he been a better physicist he would have recognized it as a form of the familiar heat exchange equation, and applied the known solution. Had he been a better mathematician, he could have solved the equation from rst principles. Certainly Merton would have known exactly what to do with the equation had he ever seen it. Perry Mehrling (2005)
Page 542
PDEs for Asian Options

Add the new variable A(t) t
0
S (u) du.
Then the value V of the Asian option satises this two-dimensional PDE:a V V 1 2 2 2V V + rS + S +S = rV. t S 2 S 2 A The terminal conditions are ( ) A V (T, S, A) = max X, 0 T ) ( A V (T, S, A) = max X , 0 T
a Kemna
for call, for put.
and Vorst (1990).
Page 543
PDEs for Asian Options (continued)

The two-dimensional PDE produces algorithms similar to that on pp. 357. But one-dimensional PDEs are available for Asian options.a For example, Ve ce r (2001) derives the following PDE for Asian calls: ( )2 2 ( ) t 1 T z 2u u t u +r 1 z + =0 2 t T z 2 z with the terminal condition u(T, z ) = max(z, 0).
a Rogers
and Shi (1995); Ve ce r (2001); Dubois and Leli` evre (2005).
Page 544
PDEs for Asian Options (concluded)

For Asian puts: (t )2 2 ( ) 2u u t u T 1z +r 1z + =0 2 t T z 2 z with the same terminal condition. One-dimensional PDEs lead to highly ecient numerical methods.
Page 545
The Hull-White Model

Hull and White (1987) postulate the following model, dS S dV = r dt + V dW1 , = v V dt + bV dW2 .
Above, V is the instantaneous variance. They assume v depends on V and t (but not S ).
Page 546
The SABR Model

Hagan, Kumar, Lesniewski, and Woodward (2002) postulate the following model, dS S dV for 0 1. A nice feature of this model is that the implied volatility surface has a compact approximate closed form. = r dt + S V dW1 , = bV dW2 ,
Page 547
The Hilliard-Schwartz Model

Hilliard and Schwartz (1996) postulate the following general model, dS S dV = r dt + f (S )V a dW1 , = (V ) dt + bV dW2 ,
for some well-behaved function f (S ) and constant a.
Page 548
The Blacher Model

Blacher (2002) postulates the following model, dS S d = = r dt + 1 + (S S0 ) + (S S0 ) ( ) dt + dW2 . [
2
dW1 ,
So the volatility follows a mean-reverting process to level .
Page 549
Hestons Stochastic-Volatility Modela

Heston assumes the stock price follows dS S dV = ( q ) dt + V dW1 , = ( V ) dt + V dW2 . (61) (62)
V is the instantaneous variance, which follows a square-root process. dW1 and dW2 have correlation . The riskless rate r is constant. It may be the most popular continuous-time stochastic-volatility model.
a Heston
(1993).
Page 550
Heston assumes the market price of risk is b2 V . So = r + b2 V . Dene

dW1 dW2
Hestons Stochastic-Volatility Model (continued)
= dW1 + b2 V dt, = dW2 + b2 V dt, = + b2 , = . + b2
dW1 and dW2 have correlation .
Page 551

Under the risk-neutral probability measure Q, both W1 and W2 are Wiener processes.
Hestons model becomes, under probability measure Q, dS S dV = (r q ) dt + V dW1 , . = ( V ) dt + V dW2
Page 552

Dene
(u, ) = exp { u(ln S + (r q ) ) [ ] d 1 ge + 2 ( u d) 2 ln 1g ( )} 2 d v ( u d) 1 e + , 1 ged (u )2 2 (u u2 ) , ( u d)/( u + d).
d g
= =
Page 553
Hestons Stochastic-Volatility Model (concluded)

The formulas area
[ C = S X (u , ) 1 1 + Re du 2 0 uSer [ ( u ) ] X ( u, ) 1 1 Xer + Re du , 2 0 u [ ( ) ] u X (u, ) 1 1 Xer Re du , 2 0 u [ ( u ) ] X (u , ) 1 1 S Re du , r 2 0 uSe
where = 1 and Re(x) denotes the real part of the complex number x.
by Mr. Chen, Chun-Ying (D95723006) on August 17, 2008 and Mr. Liou, Yan-Fu (R92723060) on August 26, 2008.
a Contributed
Page 554
Stochastic-Volatility Models and Further Extensionsa

How to explain the October 1987 crash? Stochastic-volatility models require an implausibly high-volatility level prior to and after the crash. Merton (1976) proposed jump models. Discontinuous jump models in the asset price can alleviate the problem somewhat.
a Eraker
(2004).
Page 555
Stochastic-Volatility Models and Further Extensions (continued)

But if the jump intensity is a constant, it cannot explain the tendency of large movements to cluster over time. This assumption also has no impacts on option prices. Jump-diusion models combine both. E.g., add a jump process to Eq. (61) on p. 550.
Page 556
Stochastic-Volatility Models and Further Extensions (concluded)

But they still do not adequately describe the systematic variations in option prices.a Jumps in volatility are alternatives.b E.g., add correlated jump processes to Eqs. (61) and Eq. (62) on p. 550. Such models allow high level of volatility caused by a jump to volatility.c
(2000) and Pan (2002). b Due, Pan, and Singleton (2000). c Eraker, Johnnes, and Polson (2000).
a Bates
Page 557
Complexities of Stochastic-Volatility Models

A few stochastic-volatility models suer from subexponential (c n ) tree size. Examples include the Hull-White model (1987) and the Hilliard-Schwartz model (1996).a Future research may extend this negative result to more stochastic-volatility models. We suspect many GARCH option pricing models entertain similar problems.b
(R98723059) (2012). b Chen (R95723051) (2008); Chen (R95723051), Lyuu, and Wen (D94922003) (2011).
a Chiu
Page 558
Hedging
Page 559
When Professors Scholes and Merton and I invested in warrants, Professor Merton lost the most money. And I lost the least. Fischer Black (19381995)
Page 560
Delta Hedge
The delta (hedge ratio) of a derivative f is dened as f /S . Thus f S for relatively small changes in the stock price, S . A delta-neutral portfolio is hedged as it is immunized against small changes in the stock price. A trading strategy that dynamically maintains a delta-neutral portfolio is called delta hedge.
Page 561
Delta Hedge (concluded)

Delta changes with the stock price. A delta hedge needs to be rebalanced periodically in order to maintain delta neutrality. In the limit where the portfolio is adjusted continuously, perfect hedge is achieved and the strategy becomes self-nancing.
Page 562
Implementing Delta Hedge

We want to hedge N short derivatives. Assume the stock pays no dividends. The delta-neutral portfolio maintains N shares of stock plus B borrowed dollars such that N f + N S B = 0. At next rebalancing point when the delta is , buy N ( ) shares to maintain N shares with a total borrowing of B = N S N f . Delta hedge is the discrete-time analog of the continuous-time limit and will rarely be self-nancing.
Page 563
Example
A hedger is short 10,000 European calls. = 30% and r = 6%. This calls expiration is four weeks away, its strike price is $50, and each call has a current value of f = 1.76791. As an option covers 100 shares of stock, N = 1,000,000. The trader adjusts the portfolio weekly. The calls are replicated well if the cumulative cost of trading stock is close to the call premiums FV.a
a This
example takes the replication viewpoint.
Page 564
Example (continued)
As = 0.538560, N = 538, 560 shares are purchased for a total cost of 538,560 50 = 26,928,000 dollars to make the portfolio delta-neutral. The trader nances the purchase by borrowing B = N S N f = 25,160,090 dollars net.a The portfolio has zero net value now.
takes the hedging viewpoint an alternative. See an exercise in the text.
a This
Page 565
Example (continued)
At 3 weeks to expiration, the stock price rises to $51. The new call value is f = 2.10580. So the portfolio is worth N f + 538,560 51 Be0.06/52 = 171, 622 before rebalancing.
Page 566
Example (continued)
A delta hedge does not replicate the calls perfectly; it is not self-nancing as $171,622 can be withdrawn. The magnitude of the tracking errorthe variation in the net portfolio valuecan be mitigated if adjustments are made more frequently. In fact, the tracking error over one rebalancing act is positive about 68% of the time, but its expected value is essentially zero.a The tracking error at maturity is proportional to vega.b
and Emanuel (1980). b Kamal and Derman (1999).
a Boyle
Page 567
Example (continued)
In practice tracking errors will cease to decrease beyond a certain rebalancing frequency. With a higher delta = 0.640355, the trader buys N ( ) = 101, 795 shares for $5,191,545. The number of shares is increased to N = 640, 355.
Page 568
Example (continued)
The cumulative cost is 26,928,000 e0.06/52 + 5,191,545 = 32,150,634. The portfolio is again delta-neutral.
Page 569
Option value S (1) 4 3 2 1 0 50 51 53 52 54 f (2) 1.7679 2.1058 3.3509 2.2427 4.0000 Delta (3) 0.53856 0.64036 0.85578 0.83983 1.00000
Change in delta
No. shares bought N (5)
Cost of shares (1)(6) (7) 26,928,000 5,191,545 11,417,525 829,660 8,649,450
Cumulative cost FV(8)+(7) (8) 26,928,000 32,150,634 43,605,277 42,825,960 51,524,853
(5) 0.10180 0.21542 0.01595 0.16017
(6) 538,560 101,795 215,425 15,955 160,175
The total number of shares is 1,000,000 at expiration (trading takes place at expiration, too).
Page 570
Example (concluded)
At expiration, the trader has 1,000,000 shares. They are exercised against by the in-the-money calls for $50,000,000. The trader is left with an obligation of 51,524,853 50,000,000 = 1,524,853, which represents the replication cost. Compared with the FV of the call premium, 1,767,910 e0.064/52 = 1,776,088, the net gain is 1,776,088 1,524,853 = 251,235.
Page 571
Tracking Error Revisited

Dene the dollar gamma as S 2 . The change in value of a delta-hedged long option position after a duration of t is proportional to the dollar gamma. It is about (1/2)S 2 [ (S/S )2 2 t ]. (S/S )2 is called the daily realized variance.
Page 572
Tracking Error Revisited (continued)

Let the rebalancing times be t1 , t2 , . . . , tn . Let Si = Si+1 Si . The total tracking error at expiration is about [( ] ) n 1 2 2 Si r (T ti ) Si i 2 t , e 2 Si i=0 The tracking error is path dependent.
Page 573
Tracking Error Revisited (concluded)a

The tracking error n over n rebalancing acts (such as 251,235 on p. 571) has about the same probability of being positive as being negative. Subject to certain regularity conditions, the b 2 root-mean-square tracking error E [ n ] is O(1/ n ). The root-mean-square tracking error increases with at rst and then decreases.
Kogan, and Lo (2000). b See also Grannan and Swindle (1996).
a Bertsimas,
Page 574
Delta-Gamma Hedge
Delta hedge is based on the rst-order approximation to changes in the derivative price, f , due to changes in the stock price, S . When S is not small, the second-order term, gamma 2 f /S 2 , helps (theoretically).a A delta-gamma hedge is a delta hedge that maintains zero portfolio gamma, or gamma neutrality. To meet this extra condition, one more security needs to be brought in.
a See
the numerical example on pp. 231232 of the text.
Page 575
Delta-Gamma Hedge (concluded)

Suppose we want to hedge short calls as before. A hedging call f2 is brought in. To set up a delta-gamma hedge, we solve
N f + n1 S + n2 f2 B N + n1 + n2 2 0 N + 0 + n2 2 0 = = = 0 0 0 (self-nancing), (delta neutrality), (gamma neutrality),
for n1 , n2 , and B . The gammas of the stock and bond are 0.
Page 576
Other Hedges
If volatility changes, delta-gamma hedge may not work well. An enhancement is the delta-gamma-vega hedge, which also maintains vega zero portfolio vega. To accomplish this, one more security has to be brought into the process. In practice, delta-vega hedge, which may not maintain gamma neutrality, performs better than delta hedge.
Page 577
Trees
Page 578
I love a tree more than a man. Ludwig van Beethoven (17701827) And though the holes were rather small, they had to count them all. The Beatles, A Day in the Life (1967)
Page 579
The Combinatorial Method

The combinatorial method can often cut the running time by an order of magnitude. The basic paradigm is to count the number of admissible paths that lead from the root to any terminal node. We rst used this method in the linear-time algorithm for standard European option pricing on p. 248. We will now apply it to price barrier options.
Page 580
The Reection Principlea

Imagine a particle at position (0, a) on the integral lattice that is to reach (n, b). Without loss of generality, assume a > 0 and b 0. This particles movement: (i, j ) * (i + 1, j + 1) up move S Su j (i + 1, j 1) down move S Sd
How many paths touch the x axis?

a Andr e
(1887).
Page 581
=
=
>
Page 582
The Reection Principle (continued)

For a path from (0, a) to (n, b) that touches the x axis, let J denote the rst point this happens. Reect the portion of the path from (0, a) to J . A path from (0, a) to (n, b) is constructed. It also hits the x axis at J for the rst time. The one-to-one mapping shows the number of paths from (0, a) to (n, b) that touch the x axis equals the number of paths from (0, a) to (n, b).
Page 583
The Reection Principle (concluded)

A path of this kind has (n + b + a)/2 down moves and (n b a)/2 up moves. Hence there are ( ) = ( ) (63)
n
n+a+b 2
n
nab 2
such paths for even n + a + b. (n) Convention: k = 0 for k < 0 or k > n.
Page 584
Pricing Barrier Options (Lyuu, 1998)

Focus on the down-and-in call with barrier H < X . Assume H < S without loss of generality. Dene
a h ln(X/S ) n + , 2 2 t ln(H/S ) ln (H/ (Sdn )) n = . + ln(u/d) 2 2 t = ln (X/ (Sd )) ln(u/d)
n
Suh dnh is the terminal price h is such that H that is closest to, but does not exceed H . Sua dna is the terminal price a is such that X that is closest to, but is not exceeded by X .
Page 585
Pricing Barrier Options (continued)

The true barrier is replaced by the eective barrier H in the binomial model. A process with n moves hence ends up in the money if and only if the number of up moves is at least a. The price Suk dnk is at a distance of 2k from the lowest possible price Sdn on the binomial tree. Suk dnk = Sdk dnk = Sdn2k . (64)
Page 586
5K @
5
=
D
: 0
5K = @ 5K D @
Page 587
Pricing Barrier Options (continued)

The number of paths from S to the terminal price (n) j nj Su d is j , each with probability pj (1 p)nj . With reference to p. 587, applied with a = n 2h on p. 584 by treating the Therefore, ( the reection principle can be and b = 2j 2h in Eq. (63) line as the x axis. H ) = ( )
n
n+(n2h)+(2j 2h) 2
n n 2h + j
in the process for h n/2. paths hit H
Page 588
Pricing Barrier Options (concluded)

The terminal price Suj dnj is reached by a path that hits the eective barrier with probability ( ) n pj (1 p)nj . n 2h + j The option value equals
2h (
j =a n n2h+j
p (1 p)
j
nj
Su d
j nj
) . (65)
Rn
R er /n is the riskless return per period. It implies a linear-time algorithm.a

a Lyuu
(1998).
Page 589
Convergence of BOPM
Equation (65) results in the sawtooth-like convergence shown on p. 337. The reasons are not hard to see. The true barrier most likely does not equal the eective barrier. The same holds between the strike price and the eective strike price. The issue of the strike price is less critical. But the issue of the barrier is not negligible.
Page 590
Convergence of BOPM (continued)

Convergence is actually good if we limit n to certain values191, for example. These values make the true barrier coincide with or just above one of the stock price levels, that is, H Sdj = Sej /n for some integer j . The preferred ns are thus n= 2 , j = 1, 2, 3, . . . (ln(S/H )/(j ))
Page 591
Convergence of BOPM (continued)

There is only one minor technicality left. We picked the eective barrier to be one of the n + 1 possible terminal stock prices. However, the eective barrier above, Sdj , corresponds to a terminal stock price only when n j is even.a To close this gap, we decrement n by one, if necessary, to make n j an even number.
is because j = n 2k for some k by Eq. (64) on p. 586. Of course we could have adopted the form Sdj (n j n) for the eective barrier. It makes a good exercise.
a This
Page 592
Convergence of BOPM (concluded)

The preferred ns are now if j is even n= , 1 otherwise j = 1, 2, 3, . . . , where
2
(ln(S/H )/(j ))
Evaluate pricing formula (65) on p. 589 only with the ns above.
Page 593
Down-and-in call value 5.7
5.65
5.6
5.55
5.5
500 1000 1500 2000 2500 3000 3500 #Periods
Page 594
Practical Implications
Now that barrier options can be eciently priced, we can aord to pick very large ns (p. 596). This has profound consequences (to explain later).
Page 595
Combinatorial method Value Time (milliseconds) 0.30 0.90 2.00 3.60 5.60 8.00 11.10 15.00 19.40 24.70 30.20 36.70 43.70 44.10 51.60 68.70 76.70 86.90 97.20
21 84 191 342 533 768 1047 1368 1731 2138 2587 3078 3613 4190 4809 5472 6177 6926 7717
5.507548 5.597597 5.635415 5.655812 5.652253 5.654609 5.658622 5.659711 5.659416 5.660511 5.660592 5.660099 5.660498 5.660388 5.659955 5.660122 5.659981 5.660263 5.660272
Page 596
Practical Implications (concluded)

Pricing is prohibitively time consuming when S H because n 1/ ln2 (S/H ). This is called the barrier-too-close problem. This observation is indeed true of standard quadratic-time binomial tree algorithms. But it no longer applies to linear-time algorithms (see p. 598). This binomial model is O(1/ n) convergent in general but O(1/n) convergent when the barrier is matched.a
a Lin
(R95221010) (2008) and Lin (R95221010) and Palmer (2010).
Page 597
Barrier at 95.0 n Value . . . 2743 3040 3351 3678 4021 True 2.56095 2.56065 2.56098 2.56055 2.56152 2.5615 31.1 35.5 40.1 43.8 48.1 Time n
Barrier at 99.5 Value Time n
Barrier at 99.9 Value Time
795 3184 7163 12736 19899 28656
7.47761 7.47626 7.47682 7.47661 7.47676 7.47667 7.4767
8 38 88 166 253 368
19979 79920 179819 319680 499499 719280
8.11304 8.11297 8.11300 8.11299 8.11299 8.11299 8.1130
253 1013 2200 4100 6300 8500
(All times in milliseconds.)
Page 598

Notes in Financial Engineering

Uploaded by

Copyright:

Available Formats

Notes in Financial Engineering

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Notes in Financial Engineering

Uploaded by

Copyright:

Available Formats

Product of Geometric Brownian Motion Processes

Let dY /Y dZ/Z = a dt + b dWY , = f dt + g dWZ .

Consider the Ito process U Y Z . Apply Itos lemma (Theorem 20 on p. 510):

c 2013 Prof. Yuh-Dauh Lyuu, National Taiwan University

Product of Geometric Brownian Motion Processes (continued)

c 2013 Prof. Yuh-Dauh Lyuu, National Taiwan University

Product of Geometric Brownian Motion Processes (concluded)

c 2013 Prof. Yuh-Dauh Lyuu, National Taiwan University

Quotients of Geometric Brownian Motion Processes

Keep in mind that dWY and dWZ have correlation .

14.3.6 of the textbook is erroneous.

c 2013 Prof. Yuh-Dauh Lyuu, National Taiwan University

Quotients of Geometric Brownian Motion Processes (concluded)

c 2013 Prof. Yuh-Dauh Lyuu, National Taiwan University

c 2013 Prof. Yuh-Dauh Lyuu, National Taiwan University

Forward Prices (concluded)

is also consistent with p. 464.

c 2013 Prof. Yuh-Dauh Lyuu, National Taiwan University

) 2 ( 2(tt0 ) 2(tt0 ) 1e +e Var[ x0 ], 2 ] 2 (ts) [ 2(st0 ) e 1e 2 +e

for t0 s t and X (t0 ) = x0 .

c 2013 Prof. Yuh-Dauh Lyuu, National Taiwan University

Ornstein-Uhlenbeck Process (continued)

c 2013 Prof. Yuh-Dauh Lyuu, National Taiwan University

Ornstein-Uhlenbeck Process (continued)

c 2013 Prof. Yuh-Dauh Lyuu, National Taiwan University

Ornstein-Uhlenbeck Process (concluded)

c 2013 Prof. Yuh-Dauh Lyuu, National Taiwan University

c 2013 Prof. Yuh-Dauh Lyuu, National Taiwan University

Square-Root Process (continued)

c 2013 Prof. Yuh-Dauh Lyuu, National Taiwan University

Square-Root Process (continued)

Ingersoll, and Ross (1985).

c 2013 Prof. Yuh-Dauh Lyuu, National Taiwan University

Square-Root Process (concluded)

Feller (19061970) in 1951.

c 2013 Prof. Yuh-Dauh Lyuu, National Taiwan University

Modeling Stock Prices

c 2013 Prof. Yuh-Dauh Lyuu, National Taiwan University

Modeling Stock Prices (continued)

c 2013 Prof. Yuh-Dauh Lyuu, National Taiwan University

Modeling Stock Prices (continued)

c 2013 Prof. Yuh-Dauh Lyuu, National Taiwan University

Modeling Stock Prices (continued)

(1994); Andersen and Brotherton-Ratclie (1998).

c 2013 Prof. Yuh-Dauh Lyuu, National Taiwan University

Modeling Stock Prices (concluded)

where T t and y ln(X/St ) +

c 2013 Prof. Yuh-Dauh Lyuu, National Taiwan University

Continuous-Time Derivatives Pricing

c 2013 Prof. Yuh-Dauh Lyuu, National Taiwan University

c 2013 Prof. Yuh-Dauh Lyuu, National Taiwan University

Toward the Black-Scholes Dierential Equation

c 2013 Prof. Yuh-Dauh Lyuu, National Taiwan University

c 2013 Prof. Yuh-Dauh Lyuu, National Taiwan University

Black-Scholes Dierential Equation

c 2013 Prof. Yuh-Dauh Lyuu, National Taiwan University

Black-Scholes Dierential Equation (continued)

speaking, it is not quite right (Bergman, 1982).

c 2013 Prof. Yuh-Dauh Lyuu, National Taiwan University