Reinforcement learning on a futures market simulator

Koichi Moriyama, Mitsuhiro Matsumoto, Ken Ichi Fukui, Satoshi Kurihara, Masayuki Numao

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

In recent years, market forecasting by machine learning methods has been flourishing. Most existing works use a past market data set, because they assume that each trader's individual decisions do not affect market prices at all. Meanwhile, there have been attempts to analyze economic phenomena by constructing virtual market simulators, in which human and artificial traders really make trades. Since prices in a market are, in fact, determined by every trader's decisions, a virtual market is more realistic, and the above assumption does not apply. In this work, we design several reinforcement learners on the futures market simulator U-Mart (Unreal Market as an Artificial Research Testbed) and compare our learners with the previous champions of U-Mart competitions empirically.

Original languageEnglish
Pages (from-to)1136-1153
Number of pages18
JournalJournal of Universal Computer Science
Volume14
Issue number7
Publication statusPublished - 2008 Jul 4
Externally publishedYes

Fingerprint

Reinforcement learning
Reinforcement Learning
Testbeds
Simulator
Simulators
Learning systems
Reinforcement
Economics
Testbed
Financial markets
Market
Forecasting
Machine Learning

Keywords

  • Market simulation
  • Reinforcement learning

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Moriyama, K., Matsumoto, M., Fukui, K. I., Kurihara, S., & Numao, M. (2008). Reinforcement learning on a futures market simulator. Journal of Universal Computer Science, 14(7), 1136-1153.

Reinforcement learning on a futures market simulator. / Moriyama, Koichi; Matsumoto, Mitsuhiro; Fukui, Ken Ichi; Kurihara, Satoshi; Numao, Masayuki.

In: Journal of Universal Computer Science, Vol. 14, No. 7, 04.07.2008, p. 1136-1153.

Research output: Contribution to journalArticle

Moriyama, K, Matsumoto, M, Fukui, KI, Kurihara, S & Numao, M 2008, 'Reinforcement learning on a futures market simulator', Journal of Universal Computer Science, vol. 14, no. 7, pp. 1136-1153.
Moriyama K, Matsumoto M, Fukui KI, Kurihara S, Numao M. Reinforcement learning on a futures market simulator. Journal of Universal Computer Science. 2008 Jul 4;14(7):1136-1153.
Moriyama, Koichi ; Matsumoto, Mitsuhiro ; Fukui, Ken Ichi ; Kurihara, Satoshi ; Numao, Masayuki. / Reinforcement learning on a futures market simulator. In: Journal of Universal Computer Science. 2008 ; Vol. 14, No. 7. pp. 1136-1153.
@article{befb17f56b7643a98c92ca0646d8aec6,
title = "Reinforcement learning on a futures market simulator",
abstract = "In recent years, market forecasting by machine learning methods has been flourishing. Most existing works use a past market data set, because they assume that each trader's individual decisions do not affect market prices at all. Meanwhile, there have been attempts to analyze economic phenomena by constructing virtual market simulators, in which human and artificial traders really make trades. Since prices in a market are, in fact, determined by every trader's decisions, a virtual market is more realistic, and the above assumption does not apply. In this work, we design several reinforcement learners on the futures market simulator U-Mart (Unreal Market as an Artificial Research Testbed) and compare our learners with the previous champions of U-Mart competitions empirically.",
keywords = "Market simulation, Reinforcement learning",
author = "Koichi Moriyama and Mitsuhiro Matsumoto and Fukui, {Ken Ichi} and Satoshi Kurihara and Masayuki Numao",
year = "2008",
month = "7",
day = "4",
language = "English",
volume = "14",
pages = "1136--1153",
journal = "Journal of Universal Computer Science",
issn = "0948-695X",
publisher = "Springer Verlag",
number = "7",

}

TY - JOUR

T1 - Reinforcement learning on a futures market simulator

AU - Moriyama, Koichi

AU - Matsumoto, Mitsuhiro

AU - Fukui, Ken Ichi

AU - Kurihara, Satoshi

AU - Numao, Masayuki

PY - 2008/7/4

Y1 - 2008/7/4

N2 - In recent years, market forecasting by machine learning methods has been flourishing. Most existing works use a past market data set, because they assume that each trader's individual decisions do not affect market prices at all. Meanwhile, there have been attempts to analyze economic phenomena by constructing virtual market simulators, in which human and artificial traders really make trades. Since prices in a market are, in fact, determined by every trader's decisions, a virtual market is more realistic, and the above assumption does not apply. In this work, we design several reinforcement learners on the futures market simulator U-Mart (Unreal Market as an Artificial Research Testbed) and compare our learners with the previous champions of U-Mart competitions empirically.

AB - In recent years, market forecasting by machine learning methods has been flourishing. Most existing works use a past market data set, because they assume that each trader's individual decisions do not affect market prices at all. Meanwhile, there have been attempts to analyze economic phenomena by constructing virtual market simulators, in which human and artificial traders really make trades. Since prices in a market are, in fact, determined by every trader's decisions, a virtual market is more realistic, and the above assumption does not apply. In this work, we design several reinforcement learners on the futures market simulator U-Mart (Unreal Market as an Artificial Research Testbed) and compare our learners with the previous champions of U-Mart competitions empirically.

KW - Market simulation

KW - Reinforcement learning

UR - http://www.scopus.com/inward/record.url?scp=46049110117&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=46049110117&partnerID=8YFLogxK

M3 - Article

VL - 14

SP - 1136

EP - 1153

JO - Journal of Universal Computer Science

JF - Journal of Universal Computer Science

SN - 0948-695X

IS - 7

ER -