Competing with humans at fantasy football: team formation in large partially-observable domains
Competing with humans at fantasy football: team formation in large partially-observable domains
We present the first real-world benchmark for sequentially optimal team formation, working within the framework of a class of online football prediction games known as Fantasy Football. We model the problem as a Bayesian reinforcement learning one, where the action space is exponential in the number of players and where the decision maker’s beliefs are over multiple characteristics of each footballer. We then exploit domain knowledge to construct computationally tractable solution techniques in order to build a competitive automated Fantasy Football manager. Thus, we are able to establish the baseline performance in this domain, even without complete information on footballers’ performances (accessible to human managers), showing that our agent is able to rank at around the top percentile when pitched against 2.5M human players
multi-agent systems, team formation, optimisation, sequential decision making
1394-1400
Matthews, Tim
a584a871-4934-4ac3-b5e6-ce9e6f2d880a
Ramchurn, Sarvapali
1d62ae2a-a498-444e-912d-a6082d3aaea3
Chalkiadakis, Georgios
50ef5d10-3ffe-4253-ac88-fad4004240e7
22 July 2012
Matthews, Tim
a584a871-4934-4ac3-b5e6-ce9e6f2d880a
Ramchurn, Sarvapali
1d62ae2a-a498-444e-912d-a6082d3aaea3
Chalkiadakis, Georgios
50ef5d10-3ffe-4253-ac88-fad4004240e7
Matthews, Tim, Ramchurn, Sarvapali and Chalkiadakis, Georgios
(2012)
Competing with humans at fantasy football: team formation in large partially-observable domains.
Twenty-Sixth Conference of the Association for the Advancement for Artificial Intelligence, Toronto, Canada.
22 - 26 Jul 2012.
.
Record type:
Conference or Workshop Item
(Paper)
Abstract
We present the first real-world benchmark for sequentially optimal team formation, working within the framework of a class of online football prediction games known as Fantasy Football. We model the problem as a Bayesian reinforcement learning one, where the action space is exponential in the number of players and where the decision maker’s beliefs are over multiple characteristics of each footballer. We then exploit domain knowledge to construct computationally tractable solution techniques in order to build a competitive automated Fantasy Football manager. Thus, we are able to establish the baseline performance in this domain, even without complete information on footballers’ performances (accessible to human managers), showing that our agent is able to rank at around the top percentile when pitched against 2.5M human players
Text
fantasyFootball2012cr.pdf
- Other
More information
Published date: 22 July 2012
Venue - Dates:
Twenty-Sixth Conference of the Association for the Advancement for Artificial Intelligence, Toronto, Canada, 2012-07-22 - 2012-07-26
Keywords:
multi-agent systems, team formation, optimisation, sequential decision making
Organisations:
Agents, Interactions & Complexity
Identifiers
Local EPrints ID: 340382
URI: http://eprints.soton.ac.uk/id/eprint/340382
PURE UUID: 9aa61ec1-832c-4305-8c5a-a54640568a28
Catalogue record
Date deposited: 22 Jun 2012 09:10
Last modified: 15 Mar 2024 03:22
Export record
Contributors
Author:
Tim Matthews
Author:
Sarvapali Ramchurn
Author:
Georgios Chalkiadakis
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics