UMBC Technical Report TR-02-15
UMBCTAC: A Balanced Bidding Agent
December 14, 2002
UMBCTAC is one of the top-ranking agents in the 3rd International Trading Agent Competition (TAC). A TAC game has multiple auctions running on different but interrelated resources simultaneously, and 8 trading agents will compete with each other for optimal result – making maximum profit. The spirit of simplicity and balance is used as a guideline to solve the dynamic optimization problem in the TAC game. The hotel/airline auctions and the entertainment ticket auctions are handled separately by applying an early bird heuristic. A gain-risk model is used to select a good, safe resource allocation for the clients in hotel/airline auctions. A fast and simple probabilistic approach is used to handle entertainment ticket auctions.
Downloads: 4210 downloads