Vol. 15, No. 2, 2020

Download this article
Download this article For screen
For printing
Recent Issues
Volume 18, Issue 1
Volume 17, Issue 1
Volume 16, Issue 2
Volume 16, Issue 1
Volume 15, Issue 2
Volume 15, Issue 1
Volume 14, Issue 2
Volume 14, Issue 1
Volume 13, Issue 2
Volume 13, Issue 1
Volume 12, Issue 1
Volume 11, Issue 2
Volume 11, Issue 1
Volume 10, Issue 2
Volume 10, Issue 1
Volume 9, Issue 2
Volume 9, Issue 1
Volume 8, Issue 1
Volume 7, Issue 2
Volume 7, Issue 1
Volume 6, Issue 1
Volume 5, Issue 2
Volume 5, Issue 1
Volume 4, Issue 1
Volume 3, Issue 1
Volume 2, Issue 1
Volume 1, Issue 1
The Journal
About the Journal
Editorial Board
Submission Guidelines
Submission Form
Policies for Authors
Ethics Statement
ISSN: 2157-5452 (e-only)
ISSN: 1559-3940 (print)
Author Index
To Appear
Other MSP Journals
On a computationally scalable sparse formulation of the multidimensional and nonstationary maximum entropy principle

Illia Horenko, Ganna Marchenko and Patrick Gagliardini

Vol. 15 (2020), No. 2, 129–146

Data-driven modeling and computational predictions based on the maximum entropy principle (MaxEnt principle) aim to find as simple as possible — but not simpler than necessary — models that allow one to avoid the data-overfitting problem. We derive a multivariate nonparametric and nonstationary formulation of the MaxEnt principle and show that its solution can be approximated through a numerical maximization of the sparse constrained optimization problem with regularization. Application of the resulting algorithm to popular financial benchmarks reveals memoryless models allowing for simple and qualitative descriptions of data of the major stock market indices. We compare the obtained MaxEnt models to the heteroscedastic models from computational econometrics (GARCH, GARCH-GJR, MS-GARCH, and GARCH-PML4) in terms of the model fit, complexity, and prediction quality. We compare the resulting model log-likelihoods, the values of the Bayesian information criterion, posterior model probabilities, the quality of the data autocorrelation function fits, as well as the value-at-risk prediction quality. We show that all of the seven considered major financial benchmark time series (DJI, SPX, FTSE, STOXX, SMI, HSI, and N225) are better described by conditionally memoryless MaxEnt models with nonstationary regime-switching than by the common econometric models with finite memory. This analysis also reveals a sparse network of statistically significant temporal relations for the positive and negative latent variance changes among different markets. The code is provided for open access.

machine learning, financial time series, maximum entropy, heteroscedasticity, sparsity
Mathematical Subject Classification 2010
Primary: 28D20, 37M10, 37M25, 91B84
Received: 8 October 2019
Revised: 6 May 2020
Accepted: 19 July 2020
Published: 19 November 2020
Illia Horenko
Institute of Computational Science
Facoltà di scienze informatiche
Università della Svizzera italiana
Ganna Marchenko
Institute of Computational Science
Facoltà di scienze informatiche
Università della Svizzera italiana
Patrick Gagliardini
Facoltà di scienze economiche Università della Svizzera italiana
Swiss Finance Institute