Transfer-entropy-based dynamic feature selection for evaluating bitcoin price drivers
Transfer-entropy-based dynamic feature selection for evaluating bitcoin price drivers
Despite the growing literature in cryptocurrency forecasting and their price drivers, the relationship between their price and other financial time series are an ongoing matter of debate. This study proposes a three-step methodology to cover these arguments. First, we conduct an ad-hoc analysis using transfer entropy (TE) to study the causal relationship between Bitcoin (BTC) returns and a vast array of financial time series. Then, we utilise variables with a significant amount of information flow towards BTC returns to forecast multistep-ahead BTC returns. Finally, we use explainable artificial intelligence post-hoc analysis methods to discover the contribution of each input feature to the overall forecasting. The results indicate a significant change in the information flow pattern in the first days of the COVID-19 pandemic outbreak. Additionally,
our proposed TE-based feature selection method outperforms both benchmarks, a non-feature-selection model, and backward stepwise regression.
1695-1726
Barak, Sasan
f82186de-f5b7-4224-9621-a00e7501f2c3
December 2023
Barak, Sasan
f82186de-f5b7-4224-9621-a00e7501f2c3
Barak, Sasan
(2023)
Transfer-entropy-based dynamic feature selection for evaluating bitcoin price drivers.
Journal of Futures Markets, 43 (12), .
(doi:10.1002/fut.22453).
Abstract
Despite the growing literature in cryptocurrency forecasting and their price drivers, the relationship between their price and other financial time series are an ongoing matter of debate. This study proposes a three-step methodology to cover these arguments. First, we conduct an ad-hoc analysis using transfer entropy (TE) to study the causal relationship between Bitcoin (BTC) returns and a vast array of financial time series. Then, we utilise variables with a significant amount of information flow towards BTC returns to forecast multistep-ahead BTC returns. Finally, we use explainable artificial intelligence post-hoc analysis methods to discover the contribution of each input feature to the overall forecasting. The results indicate a significant change in the information flow pattern in the first days of the COVID-19 pandemic outbreak. Additionally,
our proposed TE-based feature selection method outperforms both benchmarks, a non-feature-selection model, and backward stepwise regression.
Text
Manuscript
- Accepted Manuscript
More information
Accepted/In Press date: 15 July 2023
e-pub ahead of print date: 3 August 2023
Published date: December 2023
Identifiers
Local EPrints ID: 479918
URI: http://eprints.soton.ac.uk/id/eprint/479918
ISSN: 0270-7314
PURE UUID: 736b0712-2003-4538-afa8-614125356dc4
Catalogue record
Date deposited: 28 Jul 2023 16:50
Last modified: 30 Aug 2025 04:02
Export record
Altmetrics
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics