Supply chains are a current, challenging problem for agentbased electronic commerce. Motivated by the Trading Agent Competition Supply Chain Management (TAC SCM) scenario, we consider an individual supply chain agent as having three major subtasks: acquiring supplies, selling products, and managing its local manufacturing process. In this paper, we focus on the sales subtask. In particular, we consider the problem of finding the set of bids to customers in simultaneous reverse auctions that maximizes the agent's expected profit. The key technical challenge we address in this paper is that of determining the probability that a customer will accept a particular bid price. First, we compare several machine learning approaches to estimating the probability of bid acceptance. We then perform experiments in which we apply our learning method during actual gameplay to measure the impact on agent performance.

We present and analyze results from the 2004 Trading Agent Competition supply chain management scenario. We identify behavioral differences between the agents that contributed to their succcess in the competition. In market for components, early procurement remained an important factor despite rule changes from the previous year. We provide additional experimental evidence to confirm that this phenomenon is still rational. Strategic interations also played an important role in procurement; one agent, FreeAgent, employed a strategy designed to block other agents ’ access to suppliers at the start of the game. The different ways agents responded to this challenge were an important factor in the outcome of the tournament. In the customer sales market, average selling prices were a decisive difference between the top agents. Our analysis shows that the economic forces of supply and demand were key factors in determining overall market prices, and that some agents were more adept at finding and exploiting advantageous market conditions. 1

We describe two sales strategies used by the MinneTAC agent for the 2003 Supply Chain Management Trading Agent Competition. Both strategies estimate, as the game progresses, the probability of receiving a customer order for different offer prices. Offers are made to maximize the expected profit margin on each order. The main difference between the strategies is in how they compute the probability of receiving an order and the offer prices. The first strategy works well in high-demand games, the second was designed to improve performance in low-demand games. We analyze empirically the effect of the discount given by suppliers on orders received the first day of the game, and we show that in high-demand games there is a correlation between the offers an agent receives from suppliers the first day of the game and the agent's performance in the game.

Supply chains are a current, challenging problem for agentbased electronic commerce. Motivated by the Trading Agent Competition Supply Chain Management (TAC SCM) scenario, we consider an individual supply chain agent as having three major subtasks: acquiring supplies, selling products, and managing its local manufacturing process. In this

Abstract. The leap from decision support to autonomous systems has often raised a number of issues, namely system safety, soundness and security. Depending on the field of application, these issues can either be easily overcome or even hinder progress. In the case of Supply Chain Management (SCM), where system performance implies loss or profit, these issues are of high importance. SCM environments are often dynamic markets providing incomplete information, therefore demanding intelligent solutions which can adhere to environment rules, perceive variations, and act in order to achieve maximum revenue. Advancing on the way such autonomous solutions deal with the SCM process, we have built a robust, highly-adaptable and easily-configurable mechanism for efficiently dealing with all SCM facets, from material procurement and inventory management to goods production and shipment. Our agent has been crash-tested in one of the most challenging SCM environments, the trading agent competition SCM game and has proven capable of providing advanced SCM solutions on behalf of its owner. This paper introduces Mertacor and its main architectural primitives, provides an overview of the TAC SCM environment, and discusses Mertacor’s performance.

Abstract. This paper proposes a model for dynamic pricing that combines knowledge of production capacity and existing commitments, reasoning about uncertainty and learning of market conditions in an attempt to optimise expected profits. In particular, the changing market conditions are represented as a set of probabilities over the success rate of product prices. The dynamic pricing model is integrated into a real-time supply chain management agent using the Trading Agent Competition Supply Chain Management game as a test framework. We evaluate the agent experimentally in competition with other supply chain agents, and demonstrate the benefits of incorporating more market data into the dynamic pricing mechanism. 1

MinneTAC is an agent designed to compete in the Supply-Chain Trading Agent Competition [1]. It is also designed to support the needs of a group of researchers, each of whom is interested in different decision problems related to the competition scenario. The design of MinneTAC breaks out each basic behavior into a separate, configurable component. Dependencies between components are almost non-existent. This design allows each user to focus on a single problem and work independently, and it allows multiple user to tackle the same problem in different ways. This paper describes the design of MinneTAC and evaluates its effectiveness in support of our research agenda, and in its competitiveness in the TAC-SCM game environment. We also describe two sales strategies used by MinneTAC. Both strategies estimate, as the game progresses, the probability of receiving a customer order for different prices and compute the expected profit. Offers are made to maximize the expected profit on each order. The main difference between the two strategies is in how the probability of receiving an order and the offer prices are computed. The first strategy works well in high-demand games, the second was developed to improve performance in low-demand games. We empirically analyze the effect of the discount given by suppliers on orders received the first day of the game, and we show that in high-demand games there is a strong correlation between the offers an agent receives from suppliers on the first day of the game and the agent’s performance in the game. 1

Abstract. This paper proposes a novel method to characterize the performance of autonomous agents in the Trading Agent Competition for Supply Chain Management (TAC-SCM). We create benchmarking tools that manipulate market environments to control the conditions under which we test trading agents. Using these tools, we show how developers can inspect their agents and unveil behaviors that might otherwise have gone undiscovered. 1

Negotiating with suppliers and with customers is a key part of supply chain management. However, with recent technological advances, the mechanisms available to carry out such activities have become increasingly sophisticated, and the environment in which these activities take place has become highly dynamic. As a consequence, the overall planning of these complex trades, and the coordination of the various component activities, need to be carefully considered. In this setting, it is crucial that the intended behaviour, and through that, the desired outcomes, of these composite trading activities be expressed in a suitably precise manner. Using an approach based on the generation of negotiation plans, this paper describes (i) an approach to the specification of such complex activities, and (ii) a corresponding execution model. The proposal is illustrated and validated by means of a scenario taken from a recent trading agent competition.

for the 2003 Supply Chain Management Trading Agent Competition (TAC SCM). Both strategies estimate, as the game progresses, the probability of receiving a customer order for different prices and compute for each the expected profit. Offers are made to maximize the expected profit. The main difference between the strategies is in the way the probability of receiving an order is updated, and in the way an offer price is calculated. The first strategy works well in high demand games, but not as well in low-demand games. The second was developed to improve performance in low-demand games. We empirically analyze the effect of the discount given by suppliers on orders received the first day of the game. We show that in high-demand games there is a strong correlation between the offers an agent receives from suppliers on the first day of the game and the agent's performance in the game.