Models for Truthful Online Double Auctions
Abstract
Online double auctions (DAs) model a dynamic two-sided matching problem with private information and self-interest, and are relevant for dynamic resource and task allocation problems. We present a general method to design truthful DAs, such that no agent can benefit from misreporting its arrival time, duration, or value. The family of DAs is parameterized by a pricing rule, and includes a generalization of McAfee's truthful DA to this dynamic setting. We present an empirical study, in which we study the allocative-surplus and agent surplus for a number of different DAs. Our results illustrate that dynamic pricing rules are important to provide good market efficiency for markets with high volatility or low volume.
Cite
@article{arxiv.1207.1360,
title = {Models for Truthful Online Double Auctions},
author = {Jonathan Bredin and David C. Parkes},
journal= {arXiv preprint arXiv:1207.1360},
year = {2012}
}
Comments
Appears in Proceedings of the Twenty-First Conference on Uncertainty in Artificial Intelligence (UAI2005)