Information Technology Reference
In-Depth Information
4.
External Information
(considers
ext
1
,...,ext
n
in ESquation 1). It is possible
that signals external to the market can influence the preferences of the participants,
such as an event independent of the market causing the clients' preferences in the
market to change (e.g. unforeseen weather conditions affecting the production of
wheat and thus the market for wheat indirectly). Thus, external information can be
a valuable source of information that the agent can use to strategise in the market.
Having presented our IKB model for designing trading strategies, we now consider
a specific example of a market mechanism that has spawned a gamut of strategies, and
discuss how our model can be applied to it.
4
Applying IKB to the CDA
The CDA is a symmetric auction with multiple buyers and sellers and presently is one
of the most popular auction formats in marketplaces populated by autonomous software
agents. In CDAs, traders are allowed to submit offers to buy (bids) or to sell (asks) at
any time during the trading day. There is an outstanding bid (ask) which is the highest
bid (lowest ask) submitted in the market at any time during the auction. Furthermore,
the market clears continuously whenever a bid can be matched to an ask. Such CDAs
are widely used, indeed they are the principal financial institution for trading securities
and financial instruments (e.g. the NYSE and the NASDAQ both run variants of the
CDA). Because there is no known dominant strategy in the CDA, several researchers
have worked on competing alternatives [4,9,21], developing trading agents that have
been shown to be capable of outperforming humans in experimental settings [5]. We
now give a formalised definition of the single-unit, single-item CDA institution, whose
market state at time
t
k
is
p
M
(
t
k
)=
<g,
B
,
S
,price
(
t
k
)
,bid
(
t
k
)
,ask
(
t
k
)
>
where:
1.
g
is the good being auctioned off.
2.
B
=
b
1
,...,b
nb
is the finite set of identifiers of bidders in the market, where
nb
is
the number of current bidders.
3.
=
s
1
,...,s
ns
is the finite set of identifiers of sellers in the market, where
ns
is
the number of current sellers.
4.
price
(
t
k
) denotes the current market price of good
g
in the market. This corre-
sponds to the most recent transaction price.
5.
bid
(
t
k
) denotes the outstanding bid at time
t
k
.
6.
ask
(
t
k
) denotes the outstanding ask at time
t
k
.
S
The agent state at time
t
k
,is
p
i
(
t
k
)=
<id
i
,n
i
(
t
k
)
,
v
i
=(
v
1
,i
,...,v
n
i
(
t
k
)
,i
)
,budget
i
(
t
k
)
,
comp
i
(
t
k
)
>
where:
1.
id
i
defines the identity of the agent as either a buyer or a seller agent.
2.
n
i
(
t
k
) defines the number of items an agent wishes to buy or sell.
3.
v
i
=
is the vector of limit prices
6
ordered from highest to
lowest in the case of a bidder and vice versa in the case of a seller.
4.
budget
i
(
t
k
) is the budget available to agent
i
.
{
v
1
,i
,...,v
n
i
(
t
k
)
,i
}
6
This is the highest value at which a buyer would buy or the lowest value a seller will accept.
