![Page 1: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/1.jpg)
ArbitrageAn Undergraduate Introduction to Financial Mathematics
J. Robert Buchanan
2010
J. Robert Buchanan Arbitrage
![Page 2: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/2.jpg)
Efficient Market Hypothesis
The Efficient Market Hypothesis has many forms butessentially can be taken to mean
prices on all securities and products reflect all knowninformation,
current prices are the best unbiased estimate of the valueof the security or product,prices will adjust to any new information nearlyinstantaneously,an investor cannot outperform the market using knowninformation except through luck.
J. Robert Buchanan Arbitrage
![Page 3: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/3.jpg)
Efficient Market Hypothesis
The Efficient Market Hypothesis has many forms butessentially can be taken to mean
prices on all securities and products reflect all knowninformation,current prices are the best unbiased estimate of the valueof the security or product,
prices will adjust to any new information nearlyinstantaneously,an investor cannot outperform the market using knowninformation except through luck.
J. Robert Buchanan Arbitrage
![Page 4: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/4.jpg)
Efficient Market Hypothesis
The Efficient Market Hypothesis has many forms butessentially can be taken to mean
prices on all securities and products reflect all knowninformation,current prices are the best unbiased estimate of the valueof the security or product,prices will adjust to any new information nearlyinstantaneously,
an investor cannot outperform the market using knowninformation except through luck.
J. Robert Buchanan Arbitrage
![Page 5: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/5.jpg)
Efficient Market Hypothesis
The Efficient Market Hypothesis has many forms butessentially can be taken to mean
prices on all securities and products reflect all knowninformation,current prices are the best unbiased estimate of the valueof the security or product,prices will adjust to any new information nearlyinstantaneously,an investor cannot outperform the market using knowninformation except through luck.
J. Robert Buchanan Arbitrage
![Page 6: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/6.jpg)
Simple Arbitrage Situation
Arbitrage arises from mis-priced financial instruments.
Example
CostCo sells 100 stamps for $43.75.USPS sell 100 stamps for $44.00.
J. Robert Buchanan Arbitrage
![Page 7: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/7.jpg)
Intuitive Idea
Imagine we will bet on the outcome of an experiment.The Arbitrage Theorem states that either the probabilities ofthe outcomes are such that
all bets are fair, orthere is a betting scheme which produces a positive gainindependent of the outcome of the experiment.
J. Robert Buchanan Arbitrage
![Page 8: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/8.jpg)
Odds
The odds against an outcome X are related to probabilities ofthe outcome according to the formula:
n : m against =⇒ P(X ) =m
m + n.
The odds for an outcome X are related to probabilities of theoutcome according to the formula:
n : m in favor =⇒ P(X ) =n
m + n.
For a wager of m dollars on a event X with odds against ofn : m, if X occurs, we win n dollars, otherwise we lose ourinvestment.
J. Robert Buchanan Arbitrage
![Page 9: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/9.jpg)
Example (1 of 2)
ExampleSuppose the odds against player A defeating player B in atennis match are 3 : 1 and the odds against player B defeatingplayer A are 1 : 1.
P(A wins) = 0.25 and P(B wins) = 0.5
Determine a betting strategy which guarantees a positive netprofit regardless of the outcome of the tennis match.
J. Robert Buchanan Arbitrage
![Page 10: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/10.jpg)
Example (2 of 2)
Betting strategy: wager $1 on player A and $2 on player B.A wins: gain $3 on the first bet and lose $2 on the second,net gain of $1.B wins: lose $1 on the first bet and gain $2 on the second,net gain of $1.
There is a positive payoff no matter which player wins.
J. Robert Buchanan Arbitrage
![Page 11: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/11.jpg)
Introduction to Linear Programming
Linear programming is a branch of mathematics concernedwith optimizing a linear function of several variables subject tosome set of constraints (linear equalities or inequalities) on thevariables.
J. Robert Buchanan Arbitrage
![Page 12: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/12.jpg)
Example (1 of 3)
ExampleA bank may invest its deposits in loans which earn 6% interestper year and in the purchase of stocks which increase in valueby 13% per year. Any un-invested amount is simply held by thebank. Suppose that government regulations require that thebank invest no more than 60% of its deposits in stocks. As agood business practice the bank wishes to devote at least 25%of its deposits to loans. Determine how the bank shouldallocate its capital so as to maximize the total return on itsinvestments.
J. Robert Buchanan Arbitrage
![Page 13: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/13.jpg)
Example (2 of 3)
Assume the bank can invest a fraction x in loans andfraction y in stocks.
The total return is therefore 0.06x + 0.13y .The constraints are:
0.25 ≤ x0 ≤ y ≤ 0.60x + y ≤ 1
J. Robert Buchanan Arbitrage
![Page 14: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/14.jpg)
Example (2 of 3)
Assume the bank can invest a fraction x in loans andfraction y in stocks.The total return is therefore 0.06x + 0.13y .
The constraints are:0.25 ≤ x0 ≤ y ≤ 0.60x + y ≤ 1
J. Robert Buchanan Arbitrage
![Page 15: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/15.jpg)
Example (2 of 3)
Assume the bank can invest a fraction x in loans andfraction y in stocks.The total return is therefore 0.06x + 0.13y .The constraints are:
0.25 ≤ x0 ≤ y ≤ 0.60x + y ≤ 1
J. Robert Buchanan Arbitrage
![Page 16: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/16.jpg)
Example (3 of 3)
Feasible Region
0.06 x + 0.13 y = k
0.0 0.2 0.4 0.6 0.8 1.0
0.0
0.2
0.4
0.6
0.8
1.0
x
y
Optimal return occurs when x = 0.4 and y = 0.6.J. Robert Buchanan Arbitrage
![Page 17: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/17.jpg)
Linear Programming Notation (1 of 3)
If c and x are vectors with n components each, the notation
cT x = c1x1 + c2x2 + · · ·+ cnxn
represents a weighted sum of the components of x with theweights being the components of c.
Constraints will be expressed in the form aT x ≤ z.
aT x ≥ z ⇐⇒ (−a)T x ≤ −zaT x = z ⇐⇒ aT x ≤ z and (−a)T x ≤ −z
J. Robert Buchanan Arbitrage
![Page 18: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/18.jpg)
Linear Programming Notation (1 of 3)
If c and x are vectors with n components each, the notation
cT x = c1x1 + c2x2 + · · ·+ cnxn
represents a weighted sum of the components of x with theweights being the components of c.
Constraints will be expressed in the form aT x ≤ z.
aT x ≥ z ⇐⇒ (−a)T x ≤ −zaT x = z ⇐⇒ aT x ≤ z and (−a)T x ≤ −z
J. Robert Buchanan Arbitrage
![Page 19: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/19.jpg)
Linear Programming Notation (2 of 3)
We write u < v if ui < vi for i = 1,2, . . . ,n. Similarly foru > v,u ≤ v, andu ≥ v.
If 0 denotes the zero vector then x ≥ 0 is an example of a signconstraint.
J. Robert Buchanan Arbitrage
![Page 20: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/20.jpg)
Linear Programming Notation (2 of 3)
We write u < v if ui < vi for i = 1,2, . . . ,n. Similarly foru > v,u ≤ v, andu ≥ v.
If 0 denotes the zero vector then x ≥ 0 is an example of a signconstraint.
J. Robert Buchanan Arbitrage
![Page 21: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/21.jpg)
Linear Programming Notation (3 of 3)
The processes of maximizing and minimizing cT x areequivalent in the sense that cT x is a maximum if and only if(−c)T x is a minimum.
Suppose there are m inequality constraints:
aT1 x ≤ b1
aT2 x ≤ b2
...aT
mx ≤ bm
J. Robert Buchanan Arbitrage
![Page 22: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/22.jpg)
Linear Programming Notation (3 of 3)
The processes of maximizing and minimizing cT x areequivalent in the sense that cT x is a maximum if and only if(−c)T x is a minimum.
Suppose there are m inequality constraints:
aT1 x ≤ b1
aT2 x ≤ b2
...aT
mx ≤ bm
J. Robert Buchanan Arbitrage
![Page 23: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/23.jpg)
General Linear Program
Ax =
a11 a12 · · · a1na21 a22 · · · a2n
......
...am1 am2 · · · amn
x1x2...
xn
≤
b1b2...
bm
= b.
The general form of a linear program will be
“Maximize cT x subject to the constraints Ax ≤ b andx ≥ 0.”
J. Robert Buchanan Arbitrage
![Page 24: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/24.jpg)
Feasible Vectors and Cost Functions
DefinitionVector x is feasible if x ≥ 0 and Ax ≤ b.
DefinitionIf c is a vector of n components, then we define
cT x = c1x1 + c2x2 + · · ·+ cnxn
to be the cost function.
DefinitionVector x is an optimal solution if x is feasible and maximizesthe cost function.
J. Robert Buchanan Arbitrage
![Page 25: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/25.jpg)
Feasible Vectors and Cost Functions
DefinitionVector x is feasible if x ≥ 0 and Ax ≤ b.
DefinitionIf c is a vector of n components, then we define
cT x = c1x1 + c2x2 + · · ·+ cnxn
to be the cost function.
DefinitionVector x is an optimal solution if x is feasible and maximizesthe cost function.
J. Robert Buchanan Arbitrage
![Page 26: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/26.jpg)
Feasible Vectors and Cost Functions
DefinitionVector x is feasible if x ≥ 0 and Ax ≤ b.
DefinitionIf c is a vector of n components, then we define
cT x = c1x1 + c2x2 + · · ·+ cnxn
to be the cost function.
DefinitionVector x is an optimal solution if x is feasible and maximizesthe cost function.
J. Robert Buchanan Arbitrage
![Page 27: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/27.jpg)
Example (1 of 2)
Example
Use the notion of the intersection of planes in R3 to minimize5x1 + 4x2 + 8x3 subject to x1 + x2 + x3 = 1 and x is feasible.
J. Robert Buchanan Arbitrage
![Page 28: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/28.jpg)
Example (2 of 2)
0.0
0.5
1.0
x1
0.0
0.5
1.0
x2
0.0
0.5
1.0
x3
Constraint Set
0.0
0.5
1.0
x1
0.0
0.5
1.0
x2
0.0
0.5
1.0
x3
Constraint Set
5x1+4x2+8x3=k
J. Robert Buchanan Arbitrage
![Page 29: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/29.jpg)
Slack Variables
Many optimization problems may include inequalityconstraints.
Inequality constraints can be converted to equality constraintsby introducing slack variables. Thus
x1 + x2 + x3 ≤ 1 becomes x1 + x2 + x3 + x4 = 1,
where x4 ≥ 0 and “takes up the slack” to produce equality.
J. Robert Buchanan Arbitrage
![Page 30: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/30.jpg)
Example (1 of 2)
Example
Use the notion of the intersection of planes in R3 to minimize5x1 + 4x2 + 8x3 subject to x1 + x2 + x3 ≤ 1 and x is feasible.
J. Robert Buchanan Arbitrage
![Page 31: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/31.jpg)
Example (2 of 2)
If the constraints are x1 + x2 + x3 ≤ 1 and x feasible, then theset of points where the solution must be found would resemblea tetrahedron with vertices at (0,0,0), (1,0,0), (0,1,0), and(0,0,1).
Feasible Region
5x1+4x2+8x3=k-0.5
0.0
0.5
1.0
x1
-0.5
0.0
0.5
1.0
x2
-0.5
0.0
0.5
1.0
x3
J. Robert Buchanan Arbitrage
![Page 32: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/32.jpg)
Matrix Notation for Slack Variables
Suppose A is an m × n matrix, x is a vector of n components,and b is a vector of m components, then by augmenting x withm slack variables and A with the m ×m identity matrix theinequality constraint Ax ≤ b is equivalent to
a11 a12 · · · a1n 1 0 · · · 0a21 a22 · · · a2n 0 1 · · · 0
......
......
......
am1 am2 · · · amn 0 0 · · · 1
x1x2...
xnx̂n+1x̂n+2
...x̂n+m
=
b1b2...
bm
[A Im
] [ xx̂
]= b.
J. Robert Buchanan Arbitrage
![Page 33: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/33.jpg)
Canonical Linear Form
The general linear problem can now be stated in equivalentform
“maximize cT x subject to Ax = b and x ≥ 0”,
whereA is an m × (n + m) matrix consisting of the originalconstraint matrix augmented with the identity matrix, andx is the previous solution vector augmented with the slackvariables.
This new form of the linear problem will be called the canonicalform.
J. Robert Buchanan Arbitrage
![Page 34: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/34.jpg)
Dual Problems
For every linear programming problem of the type discussedabove, there is an associated problem known as its dual.Henceforth the original problem will be known as the primal.These paired optimization problems are related in the followingways.
Primal: Maximize cT x subject to Ax ≤ b and x ≥ 0.Dual: Minimize bT y subject to AT y ≥ c and y ≥ 0.
J. Robert Buchanan Arbitrage
![Page 35: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/35.jpg)
Observations
Primal: Maximize cT x subject to Ax ≤ b and x ≥ 0.Dual: Minimize bT y subject to AT y ≥ c and y ≥ 0.
Note:1 the process of maximization in the primal is replaced with
the process of minimization in the dual,2 the unknown of the dual is a vector y with m components,3 the vector b moves from the constraint of the primal to the
cost function of the dual,4 the vector c moves from the cost of the primal to the
constraint of the dual,5 the constraints of the dual are inequalities and there are n
of them.
J. Robert Buchanan Arbitrage
![Page 36: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/36.jpg)
Dual of the Dual
TheoremThe dual of the dual is the primal.
J. Robert Buchanan Arbitrage
![Page 37: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/37.jpg)
Proof
Starting with the dual problem,
Minimize bT y subject to AT y ≥ c and y ≥ 0.
We can re-write the dual in general form,
Maximize (−b)T y subject to (−A)T y ≤ −c and y ≥ 0.
Now the dual of this problem (i.e., the dual of the dual) is
Minimize (−c)T x subject to ((−A)T )T x ≥ −b and x ≥ 0.
This problem is logically equivalent to the problem
Maximize cT x subject to Ax ≤ b and x ≥ 0,
which is the primal problem.
J. Robert Buchanan Arbitrage
![Page 38: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/38.jpg)
Weak Duality Theorem
Theorem (Weak Duality Theorem)If x and y are the feasible solutions of the primal and dualproblems respectively, then cT x ≤ bT y. If cT x = bT y thenthese solutions are optimal for their respective problems.
J. Robert Buchanan Arbitrage
![Page 39: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/39.jpg)
Proof of Weak Duality Theorem
Feasible solutions to the primal and the dual problems mustsatisfy the constraints Ax ≤ b with x ≥ 0 (for the primalproblem) and AT y ≥ c with y ≥ 0 (for the dual). Multiply theconstraint in the dual by xT
xT AT y ≥ xT c ⇐⇒ cT x ≤ yT Ax.
Multiply the constraint in the primal by yT
yT Ax ≤ yT b = bT y.
Directions of the inequalities are preserved because x ≥ 0 andy ≥ 0. Combining these last two inequalities produces
cT x ≤ yT Ax ≤ bT y.
Therefore we have cT x ≤ bT y. If cT x = bT y then x and y mustbe optimal since no x can make cT x larger than bT y and no ycan make bT y smaller than cT x.
J. Robert Buchanan Arbitrage
![Page 40: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/40.jpg)
Example (1 of 2)
ExamplePrimal: Maximize 4x1 + 3x2 subject to x1 + x2 ≤ 2 andx1, x2 ≥ 0.Dual: Minimize 2y1 subject to y1 ≥ 3, y1 ≥ 4, and y1 ≥ 0.
J. Robert Buchanan Arbitrage
![Page 41: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/41.jpg)
Example (2 of 2)
The minimum value of y1 subject to the constraints must bey1 = 4. According to the Weak Duality Theorem then theminimum of the cost function of the primal must be at least 8.Applying the level set argument as before, the largest value of kfor which the level set 4x1 + 3x2 = k intersects the set offeasible points for the primal is k = 8.
Feasible Region
4x1+3x2=8
0.0 0.5 1.0 1.5 2.0
0.0
0.5
1.0
1.5
2.0
x1
x 2
J. Robert Buchanan Arbitrage
![Page 42: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/42.jpg)
More on Duality
TheoremOptimality in the primal and dual problems requires eitherxj = 0 or (AT y)j = cj for each j = 1, . . . ,n.
J. Robert Buchanan Arbitrage
![Page 43: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/43.jpg)
Proof
When x and y are optimal for their respective problems then
bT y = yT Ax = cT x(yT A− cT )x = 0(AT y− c)T x = 0.
Since x ≥ 0 and AT y− c ≥ 0 then vector x must be zero inevery component for which vector AT y− c is positive and viceversa.
J. Robert Buchanan Arbitrage
![Page 44: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/44.jpg)
Example (1 of 4)
Example
Primal: Maximize cT x = −3x1 + 2x2 − x3 + 3x4 subject tox ≥ 0 and
[1 1 −1 0−2 0 1 1
]x1x2x3x4
≤ [ 53
]
J. Robert Buchanan Arbitrage
![Page 45: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/45.jpg)
Example (2 of 4)
Dual: Minimize bT y = 5y1 + 3y2 subject to1 −21 0−1 10 1
[ y1y2
]≥
−32−13
J. Robert Buchanan Arbitrage
![Page 46: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/46.jpg)
Example (3 of 4)
1 2 3 4 5 6
2.0
2.5
3.0
3.5
4.0
4.5
5.0
y1
y 2
Optimal solution is at (y1, y2) = (3,3) and has value 24.J. Robert Buchanan Arbitrage
![Page 47: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/47.jpg)
Example (4 of 4)
Strict inequality is present in the second and third constraintssince
y1 = 3 > 2−y1 + y2 = 0 > −1.
Thus the second and third components of x in the primalproblem must be zero. Therefore the primal can be recast asPrimal: Maximize −3x1 + 3x4 subject to x1 ≥ 0, x4 ≥ 0 and
[1 1 −1 0−2 0 1 1
]x100x4
=
[x1
−2x1 + x4
]≤[
53
]
Thus x1 = 5 and x4 = 13, the maximum of the cost function forthe primal is 24 and it occurs at (x1, x2, x3, x4) = (5,0,0,13).
J. Robert Buchanan Arbitrage
![Page 48: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/48.jpg)
Duality Theorem
Theorem (Duality Theorem)If there is an optimal x in the primal, then there is an optimal yin the dual and the minimum of cT x equals the maximum ofyT b.
Remark: before proving the Duality Theorem we must state alemma which will be used in the proof.
Lemma (Farkas Alternative)Exactly one of the following two statements is true. Either
1 Ax ≤ b has a solution x ≥ 0, or2 AT y ≥ 0 with bT y < 0 has a solution y ≥ 0.
J. Robert Buchanan Arbitrage
![Page 49: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/49.jpg)
Duality Theorem
Theorem (Duality Theorem)If there is an optimal x in the primal, then there is an optimal yin the dual and the minimum of cT x equals the maximum ofyT b.
Remark: before proving the Duality Theorem we must state alemma which will be used in the proof.
Lemma (Farkas Alternative)Exactly one of the following two statements is true. Either
1 Ax ≤ b has a solution x ≥ 0, or2 AT y ≥ 0 with bT y < 0 has a solution y ≥ 0.
J. Robert Buchanan Arbitrage
![Page 50: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/50.jpg)
Proof (1 of 6)
Primal: Maximize cT x subject to Ax ≤ b and x ≥ 0.Dual: Minimize bT y subject to AT y ≥ c and y ≥ 0.
Assuming there are feasible solutions to each problem then wecan re-write the constraint of the dual as (−A)T y ≤ −c withy ≥ 0. Thus according to the constraint on the primal, there-written constraint on the dual, and the conclusion of theWeak Duality Theorem the following inequalities hold forx,y ≥ 0.
Ax ≤ b(−A)T y ≤ −c
cT x− bT y ≤ 0
J. Robert Buchanan Arbitrage
![Page 51: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/51.jpg)
Proof (2 of 6)
These inequalities can be written in the block matrix form A 00 −AT
cT −bT
[ xy
]≤
b−c
0
.According to the Farkas Alternative Lemma either thisinequality has a solution 〈x,y〉 ≥ 0 or the alternative
[AT 0 c
0 −A −b
] uvλ
≥ 0 and[
bT −cT 0] u
vλ
< 0
has a solution 〈u,v, λ〉 ≥ 0.
J. Robert Buchanan Arbitrage
![Page 52: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/52.jpg)
Proof (3 of 6)
We may decompose this block matrix to derive the followingsystem of inequalities:
AT u + λc ≥ 0, −Av− λb ≥ 0, bT u− cT v < 0
with u ≥ 0, v ≥ 0, and λ ≥ 0. If λ > 0 then this system ofinequalities is equivalent to the following system.
A(
1λ
v)≤ −b
AT(
1λ
u)≥ −c
−bT(
1λ
u)
> −cT(
1λ
v)
Since u ≥ 0 and v ≥ 0 the vectors 1λu ≥ 0 and 1
λv ≥ 0 as well.
J. Robert Buchanan Arbitrage
![Page 53: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/53.jpg)
Proof (4 of 6)
The first two inequalities form a primal problem and its dual.
Primal: A(
1λ
v)≤ −b
Dual: AT(
1λ
u)≥ −c
If we apply the Weak Duality Theorem, then it must be the casethat −bT ( 1
λu)≤ −cT ( 1
λv), contradicting the inequality:
−bT(
1λ
u)> −cT
(1λ
v)
Therefore we know that λ = 0.
J. Robert Buchanan Arbitrage
![Page 54: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/54.jpg)
Proof (5 of 6)
Thus the Farkas Alternative simplifies to the following system:
Av ≤ 0, AT u ≥ 0, and bT u < cT v
where u ≥ 0 and v ≥ 0. The last inequality implies thatbT u < 0 or cT v > 0. If bT u < 0 then the primal problem Ax ≤ bhas no feasible solution x ≥ 0. To see this note that togetherthe inequalities x ≥ 0, Ax ≤ b, and bT u < 0 imply that
(Ax)T ≤ bT
xT AT ≤ bT
xT(
AT u)≤ bT u < 0.
However, x ≥ 0 and AT u ≥ 0 and thus xT (AT u)≥ 0, a
contradiction.
J. Robert Buchanan Arbitrage
![Page 55: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/55.jpg)
Proof (6 of 6)
If cT v > 0 then the dual problem AT y ≥ c has no feasiblesolution y ≥ 0. To see this note that together the inequalitiesy ≥ 0, AT y ≥ c, and cT v > 0 imply that(
AT y)T
≥ cT
yT A ≥ cT
yT (Av) ≥ cT v > 0yT (−Av) < 0
However, y ≥ 0 and −Av ≥ 0 and thus yT (−Av) ≥ 0, acontradiction.
J. Robert Buchanan Arbitrage
![Page 56: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/56.jpg)
Fundamental Theorem of Finance (1 of 2)
Assumptions and background:Experiment has m possible outcomes numbered 1 throughm.We can place n wagers (numbered 1 through n) on theoutcomes.rij is the return for a unit bet on wager i ∈ {1,2, . . . ,n}when the outcome of the experiment is j ∈ {1,2, . . . ,m}.Vector x = (x1, x2, . . . , xn) is called a betting strategy.Component xi is the amount placed on wager i .Return from a betting strategy is
∑ni=1 xi rij .
J. Robert Buchanan Arbitrage
![Page 57: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/57.jpg)
Fundamental Theorem of Finance (2 of 2)
Theorem (Arbitrage Theorem)
Exactly one of the following is true: either1 there is a vector of probabilities y = 〈y1, y2, . . . , ym〉 for
which
m∑j=1
yj rij = 0, for each i = 1,2, . . . ,n, or
2 there is a betting strategy x = 〈x1, x2, . . . , xn〉 for which
n∑i=1
xi rij > 0, for each j = 1,2, . . . ,m.
J. Robert Buchanan Arbitrage
![Page 58: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/58.jpg)
Proof (1 of 4)
Let the n-tuple (x1, x2, . . . , xn) be the betting strategy.Vector x = 〈x1, x2, . . . , xn, xn+1〉 where the (n + 1)stelement is the payoff for the betting strategy.
We would like to maximize xn+1 under the constraint that∑ni=1 xi rji ≥ xn+1 for j = 1,2, . . . ,m. This is equivalent to∑ni=1 xi(−rji) + xn+1 ≤ 0 for j = 1,2, . . . ,m.
c = 〈 0, . . . ,0︸ ︷︷ ︸n components
,1〉 and b = 〈 0, . . . ,0︸ ︷︷ ︸m components
〉
The maximization problem above can be stated as “MaximizecT x = xn+1 subject to Ax ≤ b and x ≥ 0”.
J. Robert Buchanan Arbitrage
![Page 59: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/59.jpg)
Proof (2 of 4)
Ax =
−r11 −r12 · · · −r1n 1−r21 −r22 · · · −r2n 1
......
......
−rm1 −rm2 · · · −rmn 1
x1x2...
xnxn+1
≤
00...0
.
The dual of this primal problem can be stated as “MinimizebT y = 0 subject to AT y ≥ c and y ≥ 0.” The unknown y is avector of m components. Since b = 0 the cost function isminimized at 0.
J. Robert Buchanan Arbitrage
![Page 60: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/60.jpg)
Proof (3 of 4)
The dual problem is that of minimizing 0 subject to y ≥ 0 and−r11 −r21 · · · −rm1−r12 −r22 · · · −rm2
......
...−r1n −r2n · · · −rmn
1 1 · · · 1
y1y2...
ym−1ym
=
00...01
.
This is equivalent to the system of equations,∑m
j=1 yj rji = 0 fori = 1,2, . . . ,n;
∑mj=1 yj = 1, and yj ≥ 0 for j = 1,2, . . . ,m. The
dual problem is feasible with a minimum value of zero if andonly if y is a probability vector for which all bets have anexpected return of zero.
J. Robert Buchanan Arbitrage
![Page 61: Arbitrage - Millersville University - Welcome to Millersville](https://reader035.vdocuments.site/reader035/viewer/2022071600/613d23ac736caf36b759c7be/html5/thumbnails/61.jpg)
Proof (4 of 4)
Suppose the dual problem is feasible, the primal problem isalso feasible since xi = 0 for i = 1,2, . . . ,n + 1 satisfies theinequality constraints of the dual. According to the DualityTheorem then the maximum of the primal problem is zerowhich means no guaranteed profit is possible. In other words if(1) is true then (2) is false.
Now suppose the dual problem is not feasible. According to theDuality Theorem the primal problem has no optimal solution.Therefore zero is not the maximum payoff, thus there is abetting strategy which produces a positive payoff. Thus we seethat if (1) is false then (2) is true.
J. Robert Buchanan Arbitrage