Introduction to Reinforcement Learning
Introduction to Reinforcement Learning
Hi-index | 0.00 |
In this paper, we investigate the investment strategy in the artificial market called U-Mart, which is designed to provide a common test bed for researchers in the fields of economics and information sciences. UMIE is the international experiment of U-Mart as a contests of trading agents. We attended UMIE 2003 and 2004, and our agent won the championship in both experiments. We examin why this agent is strong in UMIE environment. The strategy of this agent is called “on-line learning” or “real-time learning”. Concretely, the agent exploits and forecasts futures price fluctuations by means of identifying the environment in reinforcement learning. We examined an efficiency of price forecasting in the classified environment. To examine the efficacy of it, we executed experiments 1000 times with UMIE open-type simulation standard toolkits, and we verified that forecasting futures price fluctuation in our strategy is useful for better trading.