An Analysis of the 2004 Supply Chain Management Trading Agent Competition - Agent-Mediated Electronic Commerce - page 106

Ta b l e 3 . Average total quantities ordered and prices paid for components

Agent Ordered CPU Price Other Price

FreeAgent 40199 0.60 0.88

Mr.UMBC 42557 0.61 0.71

UMTac-04 40738 0.56 0.61

Botticelli 11598 0.74 0.70

Deep Maize 26902 0.58 0.65

SouthamptonSCM 39350 0.58 0.71

Ta b l e 4 . Inventory management statistics. CPU/Other is the ratio of CPUs ordered to the number
of other components ordered. The ratio needed to produce PCs is 0.33 (1 CPU:3 Other).

Agent CPU/Other Unsold CPU Unsold Other Daily Inv. Ave. Delivery Day

FreeAgent 0.34 3516 5956 30677 95

Mr.UMBC 0.31 713 15769 48877 85

UMTac-04 0.33 1832 2588 41975 94

Botticelli 0.33 475 1497 1317 160

Deep Maize 0.40 12153 10200 31026 88

SouthamptonSCM 0.33 13651 43922 67147 69

orders and were in much the same situation as if they had ordered all of their compo-
nents on day 0 as originally planned. Botticelli made large day-0 requests like the rest
of the agents, but chose not to accept any of the offers and waited until much later in
the game to purchase supplies and start production. FreeAgent and Deep Maize did
not come back with large requests immediately, but purchased additional components
in smaller chunks throughout the rest of the game.
Additional details about overall procurement and pricing are in Table 3. Four of the
agents (including the top three) purchased approximately the same number of compo-
nents overall; Botticelli and Deep Maize purchased significantly fewer components.
Prices paid for CPUs are comparable for all agents except Botticelli , which did not pro-
cure any cheap CPUs on day 0. The prices paid for non-CPU components show more
disparity. UMTac-04 paid the lowest prices due to a large day-1 purchase. FreeAgent
paid very high prices for non-CPU components. This stands in contrast to the much
lower prices paid by Deep Maize , despite the similar approach these agents took in
purchasing additional inventory over the duration of the game. FreeAgent seems to
have disregarded the prices paid for these components, in part to compensate for its
strategic maneuver on day 0.

3.4 Inventory Management

Agents' procurement strategies had important implications for inventory management.
The two lowest-scoring agents in particular were crippled by difficulties in managing
inventory that are at least partially attributable to FreeAgent 's blocking tactic. There
are a number of striking numbers in table 4, which lists inventory management statis-
tics. The first is Botticelli 's very late average delivery date for components. This agent
effectively sat out most of the game after declining early component purchases. Equally