Introduction of the HC-LVQ artificial neural network for the optimization of Mexican financial cycle indicators and the identification of their turning points in real time

Jonathan Moisés Ramírez-Bautista; Federico Hernández-Álvarez; Jonathan Moisés Ramírez-Bautista; Federico Hernández-Álvarez

doi:10.3934/DSFE.2026011

Data Science in Finance and Economics

2026, Volume 6, Issue 2: 315-325. doi: 10.3934/DSFE.2026011

Previous Article Next Article

Research article

Introduction of the HC-LVQ artificial neural network for the optimization of Mexican financial cycle indicators and the identification of their turning points in real time

Jonathan Moisés Ramírez-Bautista ^{1,2
,
,},
Federico Hernández-Álvarez ²

1.
Faculty of Science, Universidad Nacional Autónoma de México, Mexico City, Mexico
2.
Department of Systems Engineering, Faculty of Engineering, Universidad Nacional Autónoma de México, Mexico City, Mexico

Received: 20 October 2025 Revised: 16 April 2026 Accepted: 17 April 2026 Published: 01 June 2026
JEL Codes: C610, C630, F370, G170, Y100

The High-Coverage Learning Vector Quantization Artificial Neural Network introduced in this paper is a non-parametric supervised classification machine learning algorithm closely related to the Vector Quantization Method and is based on the LVQ Artificial Neural Network presented by Kohonen in 2001. The HC-LVQ method is a novel approach for identifying turning points in real-time in the Mexican financial cycle based on the Mexican Stock Exchange (MXX) by optimizing a set of four key financial indicators from a selection of global stock market indices. It proved to be a competitive method compared to other traditional methods in identifying the turning points ahead of the Bry-Boschan algorithm (1971) by pinpointing in advance all the bear points (peaks) in the test period, from January 2006 to January 2024. Specifically, the HC-LVQ anticipated the identification of the Mexican financial cycle's bear of the Subprime crisis by 5 months, signaling it in May 2007 and the bull (trough) in August 2008. In contrast, the System of Composite Indicators Coincident and Advance of the National Institute of Statistics and Geography of Mexico identified the peak of the Subprime crisis for the Mexican business cycle in April 2008 and in June 2009 its recovery; while the NBER identified the peak of the Subprime crisis of the U.S. business cycle in December 2007 and in June 2009 its trough. The HC-LVQ identified the Chinese HANG SENG, the German DAX 40, the Brazilian BOVESPA, and the SPIPSA of Santiago de Chile as the indices' set that best helped identify the turning points of the MXX.
- financial cycles,
- risk aversion,
- financial crises,
- business cycles,
- economic crises,
- HC-LVQ artificial neural network,
- LVQ artificial neural network,
- VQ method,
- Bry-Boschan algorithm
Citation: Jonathan Moisés Ramírez-Bautista, Federico Hernández-Álvarez. Introduction of the HC-LVQ artificial neural network for the optimization of Mexican financial cycle indicators and the identification of their turning points in real time[J]. Data Science in Finance and Economics, 2026, 6(2): 315-325. doi: 10.3934/DSFE.2026011

Related Papers:

Abstract

The High-Coverage Learning Vector Quantization Artificial Neural Network introduced in this paper is a non-parametric supervised classification machine learning algorithm closely related to the Vector Quantization Method and is based on the LVQ Artificial Neural Network presented by Kohonen in 2001. The HC-LVQ method is a novel approach for identifying turning points in real-time in the Mexican financial cycle based on the Mexican Stock Exchange (MXX) by optimizing a set of four key financial indicators from a selection of global stock market indices. It proved to be a competitive method compared to other traditional methods in identifying the turning points ahead of the Bry-Boschan algorithm (1971) by pinpointing in advance all the bear points (peaks) in the test period, from January 2006 to January 2024. Specifically, the HC-LVQ anticipated the identification of the Mexican financial cycle's bear of the Subprime crisis by 5 months, signaling it in May 2007 and the bull (trough) in August 2008. In contrast, the System of Composite Indicators Coincident and Advance of the National Institute of Statistics and Geography of Mexico identified the peak of the Subprime crisis for the Mexican business cycle in April 2008 and in June 2009 its recovery; while the NBER identified the peak of the Subprime crisis of the U.S. business cycle in December 2007 and in June 2009 its trough. The HC-LVQ identified the Chinese HANG SENG, the German DAX 40, the Brazilian BOVESPA, and the SPIPSA of Santiago de Chile as the indices' set that best helped identify the turning points of the MXX.

References

[1]	Aldasoro I, Hördahl P, Schrimpf A, et al. (2025) Predicting financial market stress with machine learning. BIS Working Papers, No. 1250
[2]	Arouri M, Teulon F, Rault C (2013) Equity risk premium and regional integration. Int Rev Financ Anal 28: 79-85. https://doi.org/10.1016/j.irfa.2013.02.009 doi: 10.1016/j.irfa.2013.02.009
[3]	Bekaert G, Harvey C (2000) Foreign Speculators and Emerging Equity Markets. J Financ 55: 565–613. https://doi.org/10.1111/0022-1082.00220 doi: 10.1111/0022-1082.00220
[4]	Bry G, Boschan C (1971) Cyclical analysis of Time Series: Selected Procedures and Computer Programs. NBER Technical Report, 7–63. Available from: https://www.nber.org/books-and-chapters/cyclical-analysis-time-series-selected-procedures-and-computer-programs.
[5]	Burns AF, Mitchell WC (1946) Measuring Business Cycles. National Bureau of Economic Research, 1–22.
[6]	Claessens S, Kose MA, Terrones ME (2011a) How do business and financial cycles interact? IMF Working Paper. https://doi.org/10.1016/j.jinteco.2011.11.008 doi: 10.1016/j.jinteco.2011.11.008
[7]	Claessens S, Kose AM, Terrones ME (2011b) Financial Cycles: What? How? When? IMF Working Paper 11: 1–40. https://doi.org/10.5089/9781455227037.001 doi: 10.5089/9781455227037.001
[8]	Clayton C, Dos Santos A, Maggiori M, et al. (2023) Internationalizing like China. NBER Working Papers. https://doi.org/10.3386/w30336 doi: 10.3386/w30336
[9]	Corneli F, Ferriani F, Gazzani A (2023) Macroeconomic news, the financial cycle and the commodity cycle: The Chinese footprint. SSRN. https://doi.org/10.2139/ssrn.4451665 doi: 10.2139/ssrn.4451665
[10]	Cortina J, Martinez-Peria MS, Schmukler SL, et al., (2023) The internationalization of China's Equity Markets. IMF Working Papers. https://doi.org/10.5089/9798400233876.001 doi: 10.5089/9798400233876.001
[11]	Demuth H, Beale M (2004) Neural Network Toolbox. User's Guide MATLAB, Version 4. https://doi.org/10.1002/0470848944.hsa251
[12]	Forbes K, Rigobon R (1999) No contagion, only interdependence: measuring stock market co-movements. NBER Working Paper Series, 7267. https://doi.org/10.3386/w7267 doi: 10.3386/w7267
[13]	Grinderslev OJ, Kramp PL, Krongborg AF, et al. (2017) Financial cycles: what are they and what do they look like in Denmark? Danmarks Nationalbank Working Papers, 115: 1–53.
[14]	Giusto A, Piger J (2014) Identifying Business Cycle Turning Points in Real Time with Vector Quantization. Technical Report.
[15]	Kohonen T (2001) Self-Organizing Maps. Third Edition, 246–261. Espoo, Finland: Springer Series in Information Sciences. https://doi.org/10.1007/978-3-642-56927-2_6
[16]	Modi A, Patel B, Patel N (2010) The study on co-movement of selected stock markets. Int Res J Financ Econ 47.
[17]	Schüler YS, Hiebert PP, Peltonen TA (2020) Financial cycles: Characterisation and real-time measurement. J Int Money Financ 100: 102082. https://doi.org/10.1016/j.jimonfin.2019.102082 doi: 10.1016/j.jimonfin.2019.102082

Reader Comments

Your name:*

Email:*
© 2026 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)