Feature selection and deep neural networks for stock price direction forecasting using technical analysis indicators
This paper analyzes the factor zoo, which has theoretical and empirical implications for finance, from a machine learning perspective. More specifically, we discuss feature selection in the context of deep neural network models to predict the stock price direction. We investigated a set of 124 techn...
Main Authors: | , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Elsevier
2021-09-01
|
Series: | Machine Learning with Applications |
Subjects: | |
Online Access: | http://www.sciencedirect.com/science/article/pii/S266682702100030X |
id |
doaj-932639e9af1e45c58ea469a6b456bb54 |
---|---|
record_format |
Article |
spelling |
doaj-932639e9af1e45c58ea469a6b456bb542021-08-20T04:37:09ZengElsevierMachine Learning with Applications2666-82702021-09-015100060Feature selection and deep neural networks for stock price direction forecasting using technical analysis indicatorsYaohao Peng0Pedro Henrique Melo Albuquerque1Herbert Kimura2Cayan Atreio Portela Bárcena Saavedra3University of Brasilia, Campus Darcy Ribeiro, Brasilia, Distrito Federal, 70910–900, Brazil; Brazilian Ministry of Economy, Esplanade of Ministries, Block P, Brasilia, Distrito Federal, 70048-900, Brazil; Corresponding author at: Brazilian Ministry of Economy, Esplanade of Ministries, Block P, Brasilia, Distrito Federal, 70048-900, Brazil.University of Brasilia, Campus Darcy Ribeiro, Brasilia, Distrito Federal, 70910–900, BrazilUniversity of Brasilia, Campus Darcy Ribeiro, Brasilia, Distrito Federal, 70910–900, BrazilUniversity of Brasilia, Campus Darcy Ribeiro, Brasilia, Distrito Federal, 70910–900, BrazilThis paper analyzes the factor zoo, which has theoretical and empirical implications for finance, from a machine learning perspective. More specifically, we discuss feature selection in the context of deep neural network models to predict the stock price direction. We investigated a set of 124 technical analysis indicators used as explanatory variables in the recent literature and specialized trading websites. We applied three feature selection methods to shrink the feature set aiming to eliminate redundant information from similar indicators. Using daily data from stocks of seven global market indexes between 2008 and 2019, we tested neural networks with different settings of hidden layers and dropout rates. We compared various classification metrics, taking into account profitability and transaction costs levels to analyze economic gains. The results show that the variables were not uniformly chosen by the feature selection algorithms and that the out-of-sample accuracy rate of the prediction converged to two values — besides the 50% accuracy value that would suggest market efficiency, a “strange attractor” of 65% accuracy also was achieved consistently. We also found that the profitability of the strategies did not manage to significantly outperform the Buy-and-Hold strategy, even showing fairly large negative values for some hyperparameter combinations.http://www.sciencedirect.com/science/article/pii/S266682702100030XDeep learningTechnical analysis indicatorsTime-series forecastingMarket efficiencyTrading profitability |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Yaohao Peng Pedro Henrique Melo Albuquerque Herbert Kimura Cayan Atreio Portela Bárcena Saavedra |
spellingShingle |
Yaohao Peng Pedro Henrique Melo Albuquerque Herbert Kimura Cayan Atreio Portela Bárcena Saavedra Feature selection and deep neural networks for stock price direction forecasting using technical analysis indicators Machine Learning with Applications Deep learning Technical analysis indicators Time-series forecasting Market efficiency Trading profitability |
author_facet |
Yaohao Peng Pedro Henrique Melo Albuquerque Herbert Kimura Cayan Atreio Portela Bárcena Saavedra |
author_sort |
Yaohao Peng |
title |
Feature selection and deep neural networks for stock price direction forecasting using technical analysis indicators |
title_short |
Feature selection and deep neural networks for stock price direction forecasting using technical analysis indicators |
title_full |
Feature selection and deep neural networks for stock price direction forecasting using technical analysis indicators |
title_fullStr |
Feature selection and deep neural networks for stock price direction forecasting using technical analysis indicators |
title_full_unstemmed |
Feature selection and deep neural networks for stock price direction forecasting using technical analysis indicators |
title_sort |
feature selection and deep neural networks for stock price direction forecasting using technical analysis indicators |
publisher |
Elsevier |
series |
Machine Learning with Applications |
issn |
2666-8270 |
publishDate |
2021-09-01 |
description |
This paper analyzes the factor zoo, which has theoretical and empirical implications for finance, from a machine learning perspective. More specifically, we discuss feature selection in the context of deep neural network models to predict the stock price direction. We investigated a set of 124 technical analysis indicators used as explanatory variables in the recent literature and specialized trading websites. We applied three feature selection methods to shrink the feature set aiming to eliminate redundant information from similar indicators. Using daily data from stocks of seven global market indexes between 2008 and 2019, we tested neural networks with different settings of hidden layers and dropout rates. We compared various classification metrics, taking into account profitability and transaction costs levels to analyze economic gains. The results show that the variables were not uniformly chosen by the feature selection algorithms and that the out-of-sample accuracy rate of the prediction converged to two values — besides the 50% accuracy value that would suggest market efficiency, a “strange attractor” of 65% accuracy also was achieved consistently. We also found that the profitability of the strategies did not manage to significantly outperform the Buy-and-Hold strategy, even showing fairly large negative values for some hyperparameter combinations. |
topic |
Deep learning Technical analysis indicators Time-series forecasting Market efficiency Trading profitability |
url |
http://www.sciencedirect.com/science/article/pii/S266682702100030X |
work_keys_str_mv |
AT yaohaopeng featureselectionanddeepneuralnetworksforstockpricedirectionforecastingusingtechnicalanalysisindicators AT pedrohenriquemeloalbuquerque featureselectionanddeepneuralnetworksforstockpricedirectionforecastingusingtechnicalanalysisindicators AT herbertkimura featureselectionanddeepneuralnetworksforstockpricedirectionforecastingusingtechnicalanalysisindicators AT cayanatreioportelabarcenasaavedra featureselectionanddeepneuralnetworksforstockpricedirectionforecastingusingtechnicalanalysisindicators |
_version_ |
1721201623678582784 |