Options - University of Waterlookvetzal/ACC471/options.pdf · Exotic options so far we have...
Transcript of Options - University of Waterlookvetzal/ACC471/options.pdf · Exotic options so far we have...
Options
(Text reference: Chapters 17, 18)
� background
� option payoffs
� put-call parity
� option investment strategies
� exotic options
� factors affecting option values
� arbitrage bounds on option values
� Black-Scholes option valuation
� binomial option valuation
1
Background
� modern option trading began on the CBOE in 1973, although
financial option contracts can be traced back to ancient Greece
� a call option gives its holder the right to buy the underlying security
for a price X (known as the strike price or exercise price) on or before
an expiry date T
� a put option is similar except that the holder has the right to sell
� European options can only be exercised at T , American options can
be exercised at any time up to and including T
� many options trade on financial exchanges (options on individual
stocks or stock indexes, FX, options on futures), others are OTC
(interest rate options, exotic options) or packaged as part of other
securities
2
Option payoffs
ST
$
X
payoff at T
profit
Call Writer
ST
$
X
payoff at T
profit
Call Owner
ST
$
X
payoff at T
profit
Put Writer
ST
$
X
payoff at T
profit
Put Owner
3
� the payoff of an option if it were to be exercised is known as the
intrinsic value of the option
� prior to maturity, the option will have some additional value which is
called the time value of the option
$
SX
Call $
SX
Put
4
� options can be combined to create any desired payoff pattern
� e.g. suppose you buy a call with a strike of $95, sell 2 calls with a
strike of $100, and buy a call with a strike of $105. Your payoff is
max�ST � 95 � 0 � � 2 � max
�ST � 100 � 0 ��� max
�ST � 105 � 0 � , which
looks like:
$
ST
5
Put-call parity
� consider a portfolio A with a long position in the underlying asset (e.g. stock)
and a put option
� consider another portfolio B with a call option (same strike and maturity as
the put in A) and a risk free pure discount bond maturing at the same time as
the call with a face value equal to the strike
� the payoffs at maturity T for these portfolios are:
ST � X ST � X
A Stock ST ST
Put X ST 0
X ST
B Call 0 ST X
Bond X X
X ST
6
� since the two portfolios have the same future payoffs in every possible future
state, by no-arbitrage they must sell for the same value today:
St � Pt � Ct � PV � X �� an implication of this relationship is that given any three of stock, bond, call,
and put, we can always create the fourth synthetically
– example: suppose there is no risk free bond. We can construct one by
purchasing the stock, selling a call, and buying a put.
– example: suppose there is no put option. We can create one by buying a
bond and a call and shorting the stock.
� the expression above assumes that no dividends are paid by the stock before
the options expire. If there are dividends, then the parity relationship
becomes:
St � Pt � Ct � PV � X � � PV � dividends �� even this assumes that the options are European; more complicated arbitrage
bounds are available for American options
7
Options and investments
� options (and other derivatives such as futures or forward contracts) are highly
leveraged
– example: suppose you buy a one year call option on a stock for $4. The
stock is currently trading at $50, and the option is “at-the-money” (i.e.
strike equals current stock price). If the stock price rises 20% to $60 after
one year, your rate of return is � 10 4 ��� 4 � 150%; if the stock drops
20% to $40, your rate of return is 100%.
� there is some evidence that the options markets react faster to new
information than the stock markets
� many types of securities or contracts feature embedded options (convertibles,
savings bonds, retractables/extendibles, mortgages, mortgage-backed
securities, futures contracts, some swaps, insurance, segregated funds,
index-linked GICs, etc.) Although it may be quite difficult to value these
options, keep in mind that no-one gives away free options!
8
� investment strategies using traded options
1. bullish strategies:
– if you are optimistic that the price will rise, buy a call:
– if you are confident of a price increase, and quite sure the price won’t
fall, sell a put:
9
– if you are confident of a price increase but somewhat worried about a
price decrease, use a bull spread:
– if you are bearish in the short-term and bullish in the longer term, use
a diagonal spread:
10
2. bearish strategies
– if you are very sure the price will fall, buy a put:
– if you are quite certain the price will not rise, sell a call:
11
– if you believe the price will fall, and are fairly sure it will not rise, use
a bear spread:
– if you are bullish in the short term and bearish in the long term, use a
diagonal spread:
12
– if you own a stock but you are worried about a price fall, you can use a
put hedge (a.k.a. “protective put”):
3. Neutral strategies
– if you are confident that prices will fluctuate in a narrow range, sell a
straddle:
13
– if you expect a broader range of fluctuation, sell a strangle:
– if you are less certain about the range of fluctuation, use a long
butterfly:
14
– if your view is pessimistic in the short term but optimistic in the longer
term, use a calendar spread:
– if you own the stock but expect no movement in its price, use a
covered call:
15
4. volatile strategies
– if you expect very high volatility, buy a straddle:
– if you expect high volatility, buy a strangle:
16
– if you expect moderate volatility, use a short butterfly:
� many other combinations are possible
� important note: any strategy which involves selling American options
faces the risk of early exercise!
17
Exotic options� so far we have discussed plain vanilla puts and calls, but there is an
increasing variety of exotic options, with various different features:
– Bermudan options can be exercised at some points before expiry
– digital (a.k.a. binary) options are pure bets: if the underlying asset ends
up above the strike, the holder of a digital call receives $1, otherwise the
payoff is zero
– Asian options have payoffs which depend on the average price of the
underlying asset over time
– lookback options have payoffs which are functions of either the
maximum or minimum value of the underlying asset over time
– barrier options are like vanilla puts and calls that also depend on whether
the underlying asset ever reaches certain levels during the life of the
option
– quanto options have payoffs defined by variables in one currency but are
paid in a different currency
18
Factors affecting option values
� we will deal once again with plain vanilla contracts, though most of
these also affect other kinds of options
� in general, we have the following:
Factor Effect on call value Effect on put value
Stock price + -
Strike price - +
Stock price volatility + +
Time to expiration + +
Interest rate + -
Dividend - +
19
Arbitrage bounds on option values
� call options:
– since it is a limited liability instrument, a call option cannot have
a negative value
– at most, a call option can be worth St
– we can also show that a call must be worth at least
St � PV�dividends � � PV
�X �
20
– thus we have:
� important implication: on a non-dividend paying stock, it is never
optimal to exercise an American call option prior to maturity
21
� put options:
– cannot have a negative value
– worth at most X
– we can show that a put must be worth at least
PV � dividends � � PV � X � St :
– note that it may be optimal to exercise an American put option prior to
maturity, even if the underlying stock pays no dividends
22
Black-Scholes option valuation
� to get precise option values (instead of just arbitrage bounds), we must make
assumptions about the underlying stock’s distribution of returns; the
Black-Scholes model assumes that this distribution is lognormal
� in this case, the value of a European call option on a non-dividend paying
asset is:
Ct � St N � d1 � Xe � r�T � t � N � d2 ���
where
d1 � ln � St � X � � � r � σ2 � 2 � � T t �σ � T t
d2 � d1 σ � T t
r � continously compounded risk free interest rate
σ � standard deviation of stock’s return (continuously compounded)
N ��� � � cumulative normal distribution function
23
� the value of a European put option on a non-dividend paying asset is:
Pt � Xe � r�T � t � N � d2 � St N � d1 �
� example: St � $50, X � $50, r �� 05, T t � 0 � 5 years, σ �� 40
– then:
d1 � ln � 50 � 50 � � � � 05 �� 402 � 2 � � � 5 �� 40 � � 5 � 0 � 2298
d2 � 0 � 2298 � 40 � � 5 � 0 � 0530
– from text Table 18.2, we have:
N���
22 �� � 5871 N� � � 06 �� � 4761
N���
24 �� � 5948 N� � � 04 �� � 4841
� N���
2298 ��� � 24 � � 2298�24 � � 22 � � 5871 � N
� � � 0530 ��� � � 04 � � � � 0530 �� � 04 � � � � 06 ��� � 4761
� � 2298 � � 22�24 � � 22 � � 5948
� � � 0530 � � � � 06 �� � 04 � � � 0
�06 � � � 4841
� 590873 0�478900
24
– note that we can calculate these values directly on a spreadsheet: using
the NORMSDIST function in Excel gives N � d1 � � 0 � 590880 and
N � d2 � � 0 � 478853.
– then
Ct � 50 � � 590873 � 50e � � 05� �
5 � � � 478900 � � 6 � 1899
– using the accurate values from Excel gives Ct � 6 � 1925
– to calculate the value of a put option instead of a call, we have (using the
NORMSDIST function):
N � � 2298 � �� 409120 N � � 0530 � � � 521147
Pt � 50e � � 05� �
5 � � � 521147 � 50 � � 409120 � � 4 � 9580
– we can also calculate this using put-call parity:
Pt � Ct � Xe � r�T � t � St
� 6 � 1925 � 48 � 7655 50 � 4 � 9580
25
Binomial option valuation
� the fundamental idea underlying option pricing is replication, i.e.
finding a combination of other assets which gives the same payoff as
the option and then noting that absence of arbitrage � option price
equals cost of replicating portfolio
� the easiest context to analyze is the binomial model
� e.g. one period version:
100
130
76.92
Stock price tree Call price tree
What is the price of a call with X � 98 if r � 10% per year?
26
� form a portfolio with h shares of stock and B invested in a risk free asset. The
current value of the portfolio V � 100h � B. Pick h and B to replicate the
option:
130h � 1 � 1B � 32
76 � 92h � 1 � 1B � 0� h �� 6029 � B � 42 � 16
V � 100 � � 6029 � 42 � 16 � 18 � 13 � C0
� suppose we divide the year into two periods and the stock price tree is:
100
114.02
87.71
130
100
76.92
27
� consider the same call option:
C0
Cu
Cd
Cuu � 32
Cud � 2
Cdd � 0� note that dividing the year into n intervals produces n � 1 terminal stock
prices and n � 1 terminal option values, assuming:
1. u � d � d � u
2. option is not path-dependent
28
� Cu:
130h � 1 � 0488B � 32
100h � 1 � 0488B � 2
� h � 1, B � 93 � 44, Cu � 114 � 02 � 1 � 93 � 44 � 20 � 58
� Cd :
100h � 1 � 0488B � 2
76 � 92h � 1 � 0488B � 0
� h �� 0867, B � 6 � 36, Cd � 87 � 71 � � 0867 � 6 � 36 � 1 � 24
� C0:
114 � 02h � 1 � 0488B � 20 � 58
87 � 71h � 1 � 0488B � 1 � 24
� h �� 7348, B � 60 � 26, C0 � 100 � � 7348 � 60 � 26 � 13 � 22
29
� note that replication involves a dynamic portfolio strategy, i.e. we
have to rebalance the portfolio after six months. Is the call value
really $13.22? Only if we don’t have to add any further investment to
the replicating portfolio over time, i.e. if the replicating portfolio is
self-financing.
� e.g. suppose the stock price rises to $114.02 after six months. The
initial portfolio with 0.7348 shares and -60.26 is now worth
114 � 02�
� 7348 � � 60 � 26�1 � 0488 � � 20 � 58. The portfolio is then
rebalanced to 1 share of stock and -93.44 in the risk free asset, which
is worth 114 � 02 � 93 � 44 � 20 � 58.
� in this binomial framework, it is always possible to replicate using a
self-financing strategy
30
� more formally, let the current stock price be S, the up factor be u, the
down factor be d, and define R � 1 � r f (for the appropriate periods)
S0
Su
Sd
Suu
Sud
Sdd
� note that Su � S0 � u, Sd � S0 � d, Suu � S0 � u2, Sud � S0 � u � d,
Sdd � S0 � d2
� absence of arbitrage � u � R � d
31
� for the option:
C0
Cu
Cd
Cuu
Cud
Cdd
� Cu:
hS0u2 � RB � Cuu
hS0ud � RB � Cud
� hS0�u2 � ud � � Cuu � Cud
� h �Cuu � Cud
S0u�u � d �
32
� substituting in the replicating portfolio for h gives:
Cuu � Cud
S0u�u � d ��� S0u2 � RB � Cuu
�
�Cuu � Cud � u
u � d� RB � Cuu
� then:
B �
�Cuu �
�Cuu � Cud � u
u � d ��� 1R
�
�Cuu
�u � d � �
�Cuu � Cud � u
u � d ��� 1R
�uCud � dCuu
R�u � d �
33
� solving for Cu � hSu � B gives:
Cu �
�Cuu � Cud
S0u�u � d ����� S0u � uCud � dCuu
R�u � d �
�Cuu � Cud
u � d� uCud � dCuu
R�u � d �
�
�R � d � Cuu �
�u � R � Cud
R�u � d �
�1R � ��� R � d
u � d � Cuu ��
u � Ru � d � Cud �
� check previous example for Cu (R � 1 � 0488, u �� 1 � 3 � 1 � 1402,
d � 1 � 1 � 3 � � 8771):
h �Cuu � Cud
Su�u � d � �
32 � 2130 � 100
� 1
34
B �uCud � dCuu
R�u � d � �
1 � 1402�2 � � � 8771
�32 �
1 � 0488�1 � 1402 � � 8771 � � � 93 � 44
Cu �1R � � � R � d
u � d � Cuu ��
u � Ru � d � Cud �
�1
1 � 0488 � ��� 1 � 0488 � � 87711 � 1402 � � 8771 � 32 �
�1 � 1402 � 1 � 04881 � 1402 � � 8771 � 2 �
� 20 � 58
� note that there is nothing unique about this particular node—these
formulas apply throughout the tree
35
� we have:
S
S�
S �C
C�
C �
h �C
�
� C �
S� � S �
B �uC � � dC
�
R�u � d �
C �1R
� �R � du � d � C � �
�u � Ru � d � C � �
36
� check Cd in previous example:
87 � 71
100
76 � 92
Cd
2
0
h � 2 0100 76 � 92 � � 0867
B � 1 � 1402 � 0 � � 8771 � 2 �1 � 0488 � 1 � 1402 � 8771 � � 6 � 36
C � 11 � 0488
���1 � 0488 � 87711 � 1402 � 8771 � � 2 �
�1 � 1402 1 � 04881 � 1402 � 8771 � � 0 �
� 1 � 24
37
� continue back to the initial value:
100
114 � 02
87 � 71
Cd
20 � 58
1 � 24
h � 20 � 58 1 � 24114 � 02 87 � 71 �� 7350
B � 1 � 1402 � 1 � 24 � � 8771 � 20 � 58 �1 � 0488 � 1 � 1402 � 8771 � � 60 � 29
C � 11 � 0488
���1 � 0488 � 87711 � 1402 � 8771 � � 20 � 58 �
�1 � 1402 1 � 04881 � 1402 � 8771 � � 1 � 24 �
� 13 � 22
38
� of course, given h and B, it seems easier to simply calculate C � hS � B, but
there are some significant benefits to using the apparently more complicated
option pricing expression
� recall:
C � 1R
� �R du d � � C � �
�u Ru d � � C � �
� note that since u � R � d:
0 � R du d � 1 and 0 � u R
u d � 1
� furthermore:R du d � u R
u d � 1
� these terms can be interpreted as “probabilities”, even though they have
nothing to do with the actual probabilities of the stock price moving up or
down
39
� if these were the actual probabilities, then the expected gross return
on the stock would be:�R � du � d � u � �
u � Ru � d � d � R
i.e. 1 � r f , which would be the expected return on the stock if
everyone was risk-neutral. Hence these “probabilities” are called
“risk-neutral probabilities”.
� let:R � du � d
� π
� note that π is the risk-neutral probability that the stock price will rise
� then:
Ct �1R
�πC
�
t�
1 ��1 � π � C �t
�1 �
40
� extending to two periods:
C0
π
1 � π
Cu
π
1 � π
Cd
π
1 � π
Cuu
Cud
Cdd
Cu � 1R � πCuu � � 1 π � Cud �
Cd � 1R � πCud � � 1 π � Cdd �
C0 � 1R � πCu � � 1 π � Cd �
� 1R2
�π2Cuu � 2π � 1 π � Cud � � 1 π � 2Cdd �
41
� the expression at the bottom of slide 41 is the discounted expected value of
the terminal payoff of the option using the risk-neutral probabilities, i.e.
C0 � 1R2 Eπ � CT �
� note that nothing in the argument is specific to call options, so exactly the
same expression applies to put options or in fact any type of claim contingent
on the stock (though we may have to check for early exercise in the case of
American options)
� for the n period binomial model, we have
C0 � 1Rn Eπ � CT �
so for a call option
C0 � 1Rn
n
∑j 0
�nj � π j � 1 π � n � j max
�S0u jdn � j X � 0 �
42
where �nj � � n!
j! � n j � ! ��nj � π j � 1 π � n � j
is the risk-neutral probability of a path with j up moves and n j down
moves, and max � S0u jdn � j X � 0 � is the option payoff for such a path.
� e.g. n � 3:
S0
S0u
S0d
S0u2
S0ud
S0d2
S0u3
S0u2d
S0ud2
S0d3
43
j Description # of paths Total Probability
0 3 down moves � 30 � � 3!
0! � 1! � 1 1 � � 1 π � 3
1 2 down moves, 1 up � 31 � � 3!
1! � 2! � 3 3 � π � 1 π � 2
2 1 down move, 2 up � 32 � � 3!
2! � 1! � 3 3 � π2 � 1 π �3 3 up moves � 3
3 � � 3!3! � 0! � 1 1 � π3
� complications with path-dependency since there are a total of 2n paths
� see handout for an example with a standard European put and an Asian call
44
� returning to the two-period call option example with S0 � 100, r f � � 0488,
u � � 1 � 3, d � 1 � u, and X � 98:
S0 � 100
C0 � 13 � 22
h �� 7348
B � 60 � 26
Su � 114 � 02
Cu � 20 � 58
h � 1
B � 93 � 44
Sd � 87 � 71
Cd � 1 � 24
h �� 0867
B � 6 � 36
Sud � 100
Cud � 2
Suu � 130
Cuu � 32
Sdd � 76 � 92
Cdd � 0
45
� we also have
C0 � 1R2
�π2Cuu � 2π � 1 π � Cud � � 1 π � 2Cdd �
where π � � R d ��� � u d � � � 1 � 0488 � 8771 ��� � 1 � 1402 � 8771 � � � 6528 is
the risk-neutral probability that the stock rises. Thus
C0 � � 1 � 1 � 1 ��� 65282 � 32 � � 2 � � 6528 � � 1 � 6528 � � 2 � � � 1 � 6528 � 2 � 0 � �
� 13 � 22
� some concluding comments about option valuation
– it can be cheaper to replicate using futures rather than the underlying
stock because transactions costs are lower in futures markets
– the same type of results holds in other contexts: the condition that
risk-neutral probabilities exist and that option prices may be computed as
discounted expected values is equivalent to the absence of arbitrage. In
fact, this provides a relatively simple way to derive the Black-Scholes
formula.
46
– since we only have to compute an expected value, we can use Monte
Carlo methods to price options (these can be very efficient for
high-dimensional and/or path-dependent options, though early exercise is
difficult to handle)
– remember the underlying hedging argument. In some cases people
simply value an option using Black-Scholes without thinking about
whether it makes any sense (e.g. executive stock options, American).
– if we let r be the continuously compounded annual risk free rate, σ be the
annual standard deviation of the stock, let ∆t be the time interval (e.g.
T � 1 year, with n � 50 steps, ∆t �� 02). Then set u � eσ � ∆t , d � 1 � u,
π � � er∆t d ��� � u d � to converge to the Black-Scholes price as ∆t � 0.
– example: S0 � X � 100, r �� 08, T � 1, σ � � 3. The Black-Scholes price
of a European put is $8.02. For the binomial model:
n 10 20 40 80 160 320 640
European put 7.73 7.88 7.95 7.99 8.00 8.01 8.02
American put 8.77 8.84 8.87 8.89 8.90 8.90 8.90
47