Академический Документы
Профессиональный Документы
Культура Документы
1
QFI Core MFD Chapter 2: A Primer on the Arbitrage Theorem
Arbitrage
Arbitrage - taking simulataneous positions in different assets so that one guarantees a riskless
profit higher than the risk-free rate
S(t) will denote a vector of asset prices. Each entry of this vector is the price of a specific financial
security at some point in time.
S1 ( t )
S2 ( t )
St =
...
Sn ( t )
W represents the possible payouts in the future possible states of the world.
1 ( t )
2 ( t )
W=
...
n ( t )
D will be the matrix containing payoffs. dij will denote the number of units of account paid by one
unit of security i in state j. For now, consider three assets: risk-free T-bill, an underlying asset, and
a call option. Assume that time elapses and there are two possible future states of the world.
Then K = 2 and N = 3:
d11 d12 ... d1K
d21 d22 ... d2K (1 + r) (1 + r)
Dt = = S1 ( t + ) S2 ( t + )
.. .. .. ..
. . . .
C1 (t + ) C2 (t + )
d N1 d N2 . . . d NK
QFI Core MFD Chapter 2: A Primer on the Arbitrage Theorem
Positive constants 1 and 2 can be found such that asset prices satisfy:
1 (1 + r) (1 + r)
S(t) = S1 (t + ) S2 (t + ) 1
2
C (t) C1 (t + ) C2 (t + )
if and only if there are no-arbitrage possibilities.
Define the state prices i , which represent how much investors are willing to pay for one
unit of money in state i. If state i is realized, the investor gets one unit of money, otherwise
nothing.
1
This gives us an important equation: i i = v = 1+r . Here, I am using v to denote the PV
of one unit of money in every state.
Can define the risk-neutral probabilities through the state prices. If we normalize the state
prices so they add to one and set Qi = (1 + r )i .
Then i Qi = 1, and by the arbitrage theorem we have the strict inequality 0 < Qi < 1.
Can rearrange to see we are earning the risk-free rate under measure Q in risk-neutral
EQ (C (t+1))
pricing, 1 + r = C (t)
.
Note that these are not neccessarily true probabilities, they are just convenient for pricing.
1 P
We cannot simply say C (t) = 1+ r E (C ( t + 1)), since we have no guarantee of Qi = (1 + r ) i .
c 2014 The Infinite Actuary, LLC Page 3 of 11
QFI Core MFD Chapter 2: A Primer on the Arbitrage Theorem
Suppose at time t one has information summarized by It . A random variable Xt that for all s > 0
satisfies the equality
E P [ Xt+s | It ] = Xt
EQ [ Xt+s | It ] Xt
Notes on Martingales
This is important because asset prices are martingales in the risk-neutral world only
St + s
Must define Xt+s as the discounted future asset price (1+r ) s
Example
1 1.1 1.1
100 = 100 150 1
2
C 0 50
10% interest rate, Stock price is either $100 or $150 in the next time step. Assume we want
to value a call with strike 100.
Two equations with two unknowns and a third equation showing C = 502 . Thus if we
solve for 2 we can get the call premium.
Solving the matrix equality, 1 = .7272 and 2 = .1818.
Thus, C = i i Ci = 502 = 50 .1818 = 9.09. This is the no-arbitrage price of the call.
Note that here there is a unique state price vector.
Figure 1: MFD - Ch 2
Btu+ Btd+
1
u u d d 1
St = St + + d t St + St + + d t St +
2
Ct Ctu+ Ctd+
1+ d u
S= 1+r [ S Qup + Sd Qdown ]
1 u
C= 1+r [C Qup + C d Qdown ]
EQ [ StS+ ] = 1+r
1+d 1 + (r d )
EQ [ CtC+ ] = 1 + r
Under risk-neutral expectation, the underlying S grows below the risk free rate at r d
while the call option expected return is r.
This should make intuitive sense the holder of the underlying will receive d in dividends
to compensate for the lower growth rate. The holder of the call option does not receive the
dividends, so the expected return is simply the risk free rate.
Similar idea with foreign currencies replace d with the foreign savings interest rate r f .
c 2014 The Infinite Actuary, LLC Page 5 of 11
QFI Core MFD Chapter 2: A Primer on the Arbitrage Theorem
Continuous discounting, er
Calculate how the derivative asset price relates to the price of the underlying asset at expiration
or other boundaries
Define a portfolio
1 ( t )
2 ( t )
=
...
n (t)
such that the portfolio value at time t is St0 = in=1 Si (t)i
Definition: is an arbitrage portfolio, or simply an arbitrage, if either one of the following conditions
is satisfied:
Overview
This chapter is titled How to Delta Hedge. In reality, the focus in much narrower. We
will study how to delta hedge a trading strategy when we think that actual volatility and
implied volatility differ
Actual volatility: the amount of noise in the stock price, the coefficient of the Wiener process,
the amount of randomness that actually transpires
Notation
: actual volatility
h : volatility used for hedging (i.e. what volatility is plugged into Black-Scholes for
delta hedging)
: implied volatility
e
The topic of this chapter is what h to choose if you believe that actual volatility is different
from implied volatility
This is risky. Under the law of large numbers, you would make money. But only one stock
price path will be realized
Will start with a strategy like this in the rest of the chapter, but then delta hedge
Note that the standard deviation only depends on and not e . Also, both the expected
value and standard deviation are linear in , which is a lot of risk. Hence the delta hedging
in the following sections
c 2014 The Infinite Actuary, LLC Page 7 of 11
QFI Core PWIQF Chapter 10: How to Delta Hedge
1
2 (
2 2 )S2 i dt + (i a )(( r + D )Sdt + SdX )
e
In the case of discrete hedging, the final profit will be slightly different from the difference
in option values
The total profit is the difference between the true option value using the actual volatility in
Black-Scholes and the market price of the option
1
2 (
2 2 )S2 i dt
e
Another huge advantage is that we do not need to estimate h = e precisely. We just need
to know if it is larger or smaller than the actual volatility
The above integral is path dependent and is maximized when the option ends up at-the-
money (because is largest then)
This is a quite general formula. We can derive the two P&L formulas we have seen so far:
The daily P&L fluctations can be substantial, which introduce some risks
Need to use the value of your volatility forecast for
Sticky Strike
This makes the assumption that for each fixed option, the implied volatility over time
should be roughly constant (even if they are different for different options on the same
underlying)
This is a nice assumption to have because if you are holding an option, then you can
assume the implied volatility is a constant like we have in this chapter (e
(t) = e
)
Sticky Delta
c 2014 The Infinite Actuary, LLC Page 9 of 11
QFI Core PWIQF Chapter 10: How to Delta Hedge
This assumes that options with the same moneyness or delta have the same implied
volatility
Time-Periodic Behavior
Looking at patterns of the VIX, it seems like Monday has much larger average changes
QFI Core FAQ Q23: Jensens Inequality
Definition of Convexity:
f (x + (1 )y) f ( x ) + (1 ) f (y)
for any 0 1.
Jensens Inequality:
E( f ( x )) f ( E( x ))
Convex functions:
E( f (S))
= E( f (S + e))
= E f (S) + e f 0 (S) + 21 e2 f 00 (S) + . . .
c 2014 The Infinite Actuary, LLC Page 11 of 11