Machine Learning for Trading

19 Pages Posted: 14 Aug 2017 Last revised: 4 Dec 2017

See all articles by Gordon Ritter

Gordon Ritter

New York University (NYU) - Courant Institute of Mathematical Sciences; City University of New York - Weissman School of Arts and Sciences; Rutgers, The State University of New Jersey - Financial Statistics & Risk Management; New York University (NYU) - NYU Tandon School of Engineering

Date Written: August 8, 2017

Abstract

In multi-period trading with realistic market impact, determining the dynamic trading strategy that optimizes expected utility of final wealth is a hard problem. In this paper we show that, with an appropriate choice of the reward function, reinforcement learning techniques (specifically, Q-learning) can successfully handle the risk-averse case. We provide a proof of concept in the form of a simulated market which permits a statistical arbitrage even with trading costs. The Q-learning agent finds and exploits this arbitrage.

Keywords: Finance, Investment Analysis, Machine Learning, Portfolio Optimization

JEL Classification: G11, C61

Suggested Citation

Ritter, Gordon, Machine Learning for Trading (August 8, 2017). Available at SSRN: https://ssrn.com/abstract=3015609 or http://dx.doi.org/10.2139/ssrn.3015609

Gordon Ritter (Contact Author)

New York University (NYU) - Courant Institute of Mathematical Sciences ( email )

New York University
251 Mercer Street
New York, NY 10012
United States

City University of New York - Weissman School of Arts and Sciences ( email )

One Bernard Baruch Way
New York, NY 10010
United States

Rutgers, The State University of New Jersey - Financial Statistics & Risk Management ( email )

110 Frelinghuysen Road
479 Hill Center, Busch Campus
Piscataway, NJ 08854
United States

New York University (NYU) - NYU Tandon School of Engineering ( email )

6 MetroTech Center
Brooklyn, NY 11201
United States

Register to save articles to
your library

Register

Paper statistics

Downloads
7,735
rank
675
Abstract Views
19,796
PlumX Metrics