Learning in Repeated Auctions

Online auctions are one of the most fundamental facets of the modern economy and power an industry generating hundreds of billions of dollars a year in revenue. Online auction theory has historically focused on the question of designing the best way to sell a single item to potential buyers relying on some prior knowledge agents were assumed to have on each other. In new markets, such as online advertising, however, similar items are sold repeatedly, and agents are unaware of each other or might try to manipulate each other, making the assumption invalid. Statistical learning theory now provides tools to supplement those missing pieces of information given enough data, as agents can learn from their environment to improve their strategies. This book is a comprehensive introduction to the learning techniques in repeated auctions. It covers everything from the traditional economic study of optimal one-shot auctions, through learning optimal mechanisms from a dataset of bidders past values, to showing how strategic agents can actually manipulate repeated auctions to their own advantage. The authors explore the effects of different scenarios and assumptions throughout while remaining grounded in real-world applications. Many of the ideas and algorithms described are used every day to power the Internet economy. This book provides students, researchers and practitioners with a deep understanding of the theory of online auctions and gives practical examples of how to implement in modern-day internet systems.