Auctions between Regret-Minimizing Agents
Authors:
Open Access
- 25 April 2022
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 100-111
Abstract
We analyze a scenario in which software agents implemented as regret-minimizing algorithms engage in a repeated auction on behalf of their users. We study first-price and second-price auctions, as well as their generalized versions (e.g., as those used for ad auctions). Using both theoretical analysis and simulations, we show that, surprisingly, in second-price auctions the players have incentives to misreport their true valuations to their own learning agents, while in first-price auctions it is a dominant strategy for all players to truthfully report their valuations to their agents.Keywords
All Related Versions
Funding Information
- the European Research Council (ERC) (740282)
This publication has 40 references indexed in Scilit:
- Bounding the inefficiency of outcomes in generalized second price auctionsJournal of Economic Theory, 2015
- On Learning Algorithms for Nash Equilibria2010
- Sponsored search: an overview of the concept, history, and technologyInternational Journal of Electronic Business, 2008
- Position auctionsInternational Journal of Industrial Organization, 2007
- Learning, Regret Minimization, and Equilibria2007
- Adaptive and Self-Confident On-Line Learning AlgorithmsJournal of Computer and System Sciences, 2002
- A Simple Adaptive Procedure Leading to Correlated EquilibriumEconometrica, 2000
- A sublinear-time randomized approximation algorithm for matrix gamesOperations Research Letters, 1995
- COUNTERSPECULATION, AUCTIONS, AND COMPETITIVE SEALED TENDERSThe Journal of Finance, 1961
- An Iterative Method of Solving a GameAnnals of Mathematics, 1951