Catálogo de publicaciones - libros

This paper describes the design and evaluation of SouthamptonSCM, a finalist in the 2004 International Trading Agent Supply Chain Management Competition (TAC SCM). In particular, we focus on the way in which our agent sets its prices according to the prevailing market situation and its own inventory level (because this adaptivity and flexibility are the key components of its success). Specifically, we analyse our pricing model’s performance both in the actual competition and in controlled experiments. Through this evaluation, we show that SouthamptonSCM performs well across a broad range of environments.

We systematically explore a range of variations of our TAC travel-shopping agent, Walverine. The space of strategies is defined by settings to behavioral parameter values. Our empirical game-theoretic analysis is facilitated by approximating games through hierarchical reduction methods. This approach generated a small set of candidates for the version to run in the TAC-05 tournament. We selected among these based on performance in preliminary rounds, ultimately identifying a successful strategy for Walverine 2005.

In this paper, we present a novel multi-layered framework for designing strategies for trading agents. The objective of this work is to provide a framework that will assist strategy designers with the different aspects involved in designing a strategy. At present, such strategies are typically designed in an ad-hoc and intuitive manner with little regard for discerning best practice or attaining re-usability in the design process. Given this, our aim is to put such developments on a more systematic engineering footing. After we describe our framework, we then go on to illustrate how it can be used to design strategies for a particular type of market mechanism (namely the Continuous Double Auction), and how it was used to design a novel strategy for the Travel Game of the International Trading Agent Competition.

The Sample Average Approximation (SAA) method is a technique for approximating solutions to stochastic programs. Here, we attempt to scale up the SAA method to harder problems than those previously studied. We argue that to apply the SAA method effectively, there are three parameters to optimize: the number of evaluations, the number of scenarios, and the number of candidate solutions. We propose an experimental methodology for finding the optimal settings of these parameters given fixed time and space constraints. We apply our methodology to two large-scale stochastic optimization problems that arise in the context of the annual Trading Agent Competition. Both problems are expressed as integer linear programs and solved using CPLEX. Runtime increases linearly with the number of scenarios in one of the problems, and exponentially in the other. We find that, in the former problem, maximizing the number of scenarios yields the best solution, while in the latter problem, it is necessary to evaluate multiple candidate solutions to find the best solution, since increasing the number of scenarios becomes expensive very quickly.

Market theory is often concerned only with centralised markets. In this paper, we consider a market that is distributed over a network, allowing us to characterise spatially (or temporally) segregated markets. The effect of this modification on the behaviour of a market populated by simple trading agents was examined. It was demonstrated that an agent’s ability to identify the optimum market price is positively correlated with its network connectivity. A better connected agent receives more information and, as a result, is better able to judge the market state. The ZIP trading agent algorithm is modified in light of this result. Simulations reveal that trading agents which take account of the quality of the information that they receive are better able to identify the optimum price within a market.

Mechanisms that aggregate the possibly conflicting preferences of individual agents are studied extensively in economics, operations research, and lately computer science. Perhaps surprisingly, the classic literature assumes to act selfishly, possibly untruthfully, if it is to their advantage, whereas the is usually assumed to be honest and trustworthy. We argue that cryptography offers various concepts and building blocks to ensure the secure, , correct and private, execution of mechanisms. We propose models with and without a center that guarantee correctness and preserve the privacy of preferences relying on diverse assumptions such as the trustworthiness of the center or the hardness of computation. The decentralized model in which agents jointly “emulate” a virtual mechanism center is particularly interesting for two reasons. For one, it provides privacy without relying on a trusted third-party. Second, it enables the provably correct execution of mechanisms (which is not the case in the centralized model). We furthermore point out how untruthful and multi-step mechanisms can improve privacy. In particular, we show that the fully private emulation of a preference elicitor can result in unconditional privacy of a (non-empty) subset of preferences.