Portfolio management via two-stage deep learning with a joint cost

Hyungbin Yun, Minhyeok Lee, Yeong Seon Kang, Junhee Seok

Research output: Contribution to journalArticle

Abstract

Portfolio management is a series of processes that maximize returns and minimize risk by allocating assets efficiently. Along with the developments in machine learning technology, it has been studied to apply machine learning methods to prediction-based portfolio management. However, such methods have a few limitations. First, they do not consider the relations between assets for the prediction. In addition, the studies commonly focus on the prediction accuracy, neglecting the construction of portfolios. Furthermore, the methods have usually been evaluated with index data, which hardly represent actual prices to buy or sell an asset. To overcome these problems, Exchange Traded Funds (ETFs) are employed for base assets for the evaluation, and we propose a two-stage deep learning framework, called Grouped-ETFs Model (GEM), with a joint cost function. The GEM is designed to learn the features of inter-asset and groups in each stage. Also, the proposed joint cost can consider relative returns for the training while the relative returns are a crucial factor to construct a portfolio. The results of a rigorous evaluation with global ETF data indicate that the proposed GEM with the joint cost outperforms the equally weighted portfolio and the ordinary deep learning model by 33.7% and 30.1%, respectively. An additional experiment using sector ETFs verifies the generality of the proposed model where the results accord with those of the previous experiment.

Original languageEnglish
Article number113041
JournalExpert Systems with Applications
Volume143
DOIs
Publication statusPublished - 2020 Apr 1

Fingerprint

Costs
Learning systems
Cost functions
Experiments
Deep learning

Keywords

  • Deep learning
  • Joint cost function
  • Long short-term memory
  • Portfolio management

ASJC Scopus subject areas

  • Engineering(all)
  • Computer Science Applications
  • Artificial Intelligence

Cite this

Portfolio management via two-stage deep learning with a joint cost. / Yun, Hyungbin; Lee, Minhyeok; Kang, Yeong Seon; Seok, Junhee.

In: Expert Systems with Applications, Vol. 143, 113041, 01.04.2020.

Research output: Contribution to journalArticle

Yun, Hyungbin ; Lee, Minhyeok ; Kang, Yeong Seon ; Seok, Junhee. / Portfolio management via two-stage deep learning with a joint cost. In: Expert Systems with Applications. 2020 ; Vol. 143.
@article{77238a4d85bd454ea67f73335568aeb7,
title = "Portfolio management via two-stage deep learning with a joint cost",
abstract = "Portfolio management is a series of processes that maximize returns and minimize risk by allocating assets efficiently. Along with the developments in machine learning technology, it has been studied to apply machine learning methods to prediction-based portfolio management. However, such methods have a few limitations. First, they do not consider the relations between assets for the prediction. In addition, the studies commonly focus on the prediction accuracy, neglecting the construction of portfolios. Furthermore, the methods have usually been evaluated with index data, which hardly represent actual prices to buy or sell an asset. To overcome these problems, Exchange Traded Funds (ETFs) are employed for base assets for the evaluation, and we propose a two-stage deep learning framework, called Grouped-ETFs Model (GEM), with a joint cost function. The GEM is designed to learn the features of inter-asset and groups in each stage. Also, the proposed joint cost can consider relative returns for the training while the relative returns are a crucial factor to construct a portfolio. The results of a rigorous evaluation with global ETF data indicate that the proposed GEM with the joint cost outperforms the equally weighted portfolio and the ordinary deep learning model by 33.7{\%} and 30.1{\%}, respectively. An additional experiment using sector ETFs verifies the generality of the proposed model where the results accord with those of the previous experiment.",
keywords = "Deep learning, Joint cost function, Long short-term memory, Portfolio management",
author = "Hyungbin Yun and Minhyeok Lee and Kang, {Yeong Seon} and Junhee Seok",
year = "2020",
month = "4",
day = "1",
doi = "10.1016/j.eswa.2019.113041",
language = "English",
volume = "143",
journal = "Expert Systems with Applications",
issn = "0957-4174",
publisher = "Elsevier Limited",

}

TY - JOUR

T1 - Portfolio management via two-stage deep learning with a joint cost

AU - Yun, Hyungbin

AU - Lee, Minhyeok

AU - Kang, Yeong Seon

AU - Seok, Junhee

PY - 2020/4/1

Y1 - 2020/4/1

N2 - Portfolio management is a series of processes that maximize returns and minimize risk by allocating assets efficiently. Along with the developments in machine learning technology, it has been studied to apply machine learning methods to prediction-based portfolio management. However, such methods have a few limitations. First, they do not consider the relations between assets for the prediction. In addition, the studies commonly focus on the prediction accuracy, neglecting the construction of portfolios. Furthermore, the methods have usually been evaluated with index data, which hardly represent actual prices to buy or sell an asset. To overcome these problems, Exchange Traded Funds (ETFs) are employed for base assets for the evaluation, and we propose a two-stage deep learning framework, called Grouped-ETFs Model (GEM), with a joint cost function. The GEM is designed to learn the features of inter-asset and groups in each stage. Also, the proposed joint cost can consider relative returns for the training while the relative returns are a crucial factor to construct a portfolio. The results of a rigorous evaluation with global ETF data indicate that the proposed GEM with the joint cost outperforms the equally weighted portfolio and the ordinary deep learning model by 33.7% and 30.1%, respectively. An additional experiment using sector ETFs verifies the generality of the proposed model where the results accord with those of the previous experiment.

AB - Portfolio management is a series of processes that maximize returns and minimize risk by allocating assets efficiently. Along with the developments in machine learning technology, it has been studied to apply machine learning methods to prediction-based portfolio management. However, such methods have a few limitations. First, they do not consider the relations between assets for the prediction. In addition, the studies commonly focus on the prediction accuracy, neglecting the construction of portfolios. Furthermore, the methods have usually been evaluated with index data, which hardly represent actual prices to buy or sell an asset. To overcome these problems, Exchange Traded Funds (ETFs) are employed for base assets for the evaluation, and we propose a two-stage deep learning framework, called Grouped-ETFs Model (GEM), with a joint cost function. The GEM is designed to learn the features of inter-asset and groups in each stage. Also, the proposed joint cost can consider relative returns for the training while the relative returns are a crucial factor to construct a portfolio. The results of a rigorous evaluation with global ETF data indicate that the proposed GEM with the joint cost outperforms the equally weighted portfolio and the ordinary deep learning model by 33.7% and 30.1%, respectively. An additional experiment using sector ETFs verifies the generality of the proposed model where the results accord with those of the previous experiment.

KW - Deep learning

KW - Joint cost function

KW - Long short-term memory

KW - Portfolio management

UR - http://www.scopus.com/inward/record.url?scp=85074157575&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85074157575&partnerID=8YFLogxK

U2 - 10.1016/j.eswa.2019.113041

DO - 10.1016/j.eswa.2019.113041

M3 - Article

AN - SCOPUS:85074157575

VL - 143

JO - Expert Systems with Applications

JF - Expert Systems with Applications

SN - 0957-4174

M1 - 113041

ER -