Forecasting Financial Time Series through Causal and Dilated Convolutional Neural Networks

In this paper, predictions of future price movements of a major American stock index were made by analyzing past movements of the same and other correlated indices. A model that has shown very good results in audio and speech generation was modified to suit the analysis of financial data and was the...

Full description

Bibliographic Details
Main Authors:	Lukas Börjesson, Martin Singull
Format:	Article
Language:	English
Published:	MDPI AG 2020-09-01
Series:	Entropy
Subjects:	deep learning financial time series causal and dilated convolutional neural networks
Online Access:	https://www.mdpi.com/1099-4300/22/10/1094

id	doaj-be83060fc2dc41e2a40c52cba2fa97e3
record_format	Article
spelling	doaj-be83060fc2dc41e2a40c52cba2fa97e32020-11-25T02:32:50ZengMDPI AGEntropy1099-43002020-09-01221094109410.3390/e22101094Forecasting Financial Time Series through Causal and Dilated Convolutional Neural NetworksLukas Börjesson0Martin Singull1Department of Mathematics, Linköping University, 581 83 Linköping, SwedenDepartment of Mathematics, Linköping University, 581 83 Linköping, SwedenIn this paper, predictions of future price movements of a major American stock index were made by analyzing past movements of the same and other correlated indices. A model that has shown very good results in audio and speech generation was modified to suit the analysis of financial data and was then compared to a base model, restricted by assumptions made for an efficient market. The performance of any model, trained by looking at past observations, is heavily influenced by how the division of the data into train, validation and test sets is made. This is further exaggerated by the temporal structure of the financial data, which means that the causal relationship between the predictors and the response is dependent on time. The complexity of the financial system further increases the struggle to make accurate predictions, but the model suggested here was still able to outperform the naive base model by more than 20% and 37%, respectively, when predicting the next day’s closing price and the next day’s trend.https://www.mdpi.com/1099-4300/22/10/1094deep learningfinancial time seriescausal and dilated convolutional neural networks
collection	DOAJ
language	English
format	Article
sources	DOAJ
author	Lukas Börjesson Martin Singull
spellingShingle	Lukas Börjesson Martin Singull Forecasting Financial Time Series through Causal and Dilated Convolutional Neural Networks Entropy deep learning financial time series causal and dilated convolutional neural networks
author_facet	Lukas Börjesson Martin Singull
author_sort	Lukas Börjesson
title	Forecasting Financial Time Series through Causal and Dilated Convolutional Neural Networks
title_short	Forecasting Financial Time Series through Causal and Dilated Convolutional Neural Networks
title_full	Forecasting Financial Time Series through Causal and Dilated Convolutional Neural Networks
title_fullStr	Forecasting Financial Time Series through Causal and Dilated Convolutional Neural Networks
title_full_unstemmed	Forecasting Financial Time Series through Causal and Dilated Convolutional Neural Networks
title_sort	forecasting financial time series through causal and dilated convolutional neural networks
publisher	MDPI AG
series	Entropy
issn	1099-4300
publishDate	2020-09-01
description	In this paper, predictions of future price movements of a major American stock index were made by analyzing past movements of the same and other correlated indices. A model that has shown very good results in audio and speech generation was modified to suit the analysis of financial data and was then compared to a base model, restricted by assumptions made for an efficient market. The performance of any model, trained by looking at past observations, is heavily influenced by how the division of the data into train, validation and test sets is made. This is further exaggerated by the temporal structure of the financial data, which means that the causal relationship between the predictors and the response is dependent on time. The complexity of the financial system further increases the struggle to make accurate predictions, but the model suggested here was still able to outperform the naive base model by more than 20% and 37%, respectively, when predicting the next day’s closing price and the next day’s trend.
topic	deep learning financial time series causal and dilated convolutional neural networks
url	https://www.mdpi.com/1099-4300/22/10/1094
work_keys_str_mv	AT lukasborjesson forecastingfinancialtimeseriesthroughcausalanddilatedconvolutionalneuralnetworks AT martinsingull forecastingfinancialtimeseriesthroughcausalanddilatedconvolutionalneuralnetworks
_version_	1724817332365164544

Forecasting Financial Time Series through Causal and Dilated Convolutional Neural Networks

Similar Items