Javascript must be enabled for the correct page display

Applying Echo State Networks with Reinforcement Learning to the Foreign Exchange Market

Steeg, M. van de (2017) Applying Echo State Networks with Reinforcement Learning to the Foreign Exchange Market. Master's Thesis / Essay, Artificial Intelligence.

[img]
Preview
Text
Thesis_Michiel-van-de-Steeg.pdf - Published Version

Download (2MB) | Preview
[img] Text
Toestemming.pdf - Other
Restricted to Backend only

Download (79kB)

Abstract

The foreign exchange market (forex) is the market in which currencies are traded with each other. The exchange rate is driven by supply and demand, which in turn is driven by factors such as inflation, interest rates, political and economic stability, and market sentiment, and is largely stochastic. Traders attempt to exchange currencies at the right time to make profit, and they use many types of analyses to do so. In this study, we use the forex as a testbed for echo state networks (ESN), a type of recurrent neural network. We looked at how well we can use the past to predict the change in exchange rate over the next hour, and we simulate trading based on these predictions. The ESN could not predict the change in exchange rate over the next hour better than a benchmark, nor better than random in most cases, and as such it also traded poorly. However, when combining the rich temporal representation of the ESN with Q-learning, a form of reinforcement learning, a more flexible trading agent was able to make significant profit in a simulation, despite trading costs. Author's note: due to an error in the experiments the results from chapter 3 were invalidated

Item Type: Thesis (Master's Thesis / Essay)
Degree programme: Artificial Intelligence
Thesis type: Master's Thesis / Essay
Language: English
Date Deposited: 15 Feb 2018 08:30
Last Modified: 15 Feb 2018 08:30
URI: http://fse.studenttheses.ub.rug.nl/id/eprint/15472

Actions (login required)

View Item View Item