14-Arbitrage Page 396 Wednesday, February 4, 2004 1:08 PM
396 The Mathematics of Financial Modeling and Investment Management
The finite number of states represents uncertainty. There is uncer-
tainty because the world can be in any of the M states. At time 0 it is not
known in what state the world will be at time 1. Uncertainty is quanti-
fied by probabilities but a lot of arbitrage pricing theory can be devel-
oped without any reference to probabilities. Suppose there are N
securities. Each security i pays dij number of dollars (or of any other
unit of account) in each state of the world j. The payoff of each security
need not be a positive number. For instance, a derivative instrument
might have negative payoffs in some states of the world. Therefore, in a
one-period setting, the securities are formally represented by an N×M
matrix D = {dij} where the dij entry is the payoff of security i in state j.
Recall from Chapter 5 that the matrix D can also be written as a set of
N row vectors:
d 1
D = · , di =
dN
di 1 · diM
where the M-vector di represents the payoffs of security i in each of the
M states.
Each security is characterized by a price S. Therefore, the set of N
securities is characterized by an N-vector S and an N×M matrix D. Sup-
pose, for instance, there are two states and three securities. Then the
three securities are represented by
S 1
S = S 2 , D =
S 3
d 11 d 12
d 21 d 22
d 31 d 32
Every row of the D matrix represents one security, every column one
state. Note that in a one-period setting, prices are defined at time 0
while payoffs are defined at time 1. There is no payoff at time 0 and
there is no price at time 1. A portfolio is represented by a N-vector of
weights θθθθ. In our example of a market with two states and three securi-
ties, a portfolio is a 3-vector:
θθθθ=
θ 1
θ 2
θ 3