ABSTRACT
Although advanced statistical models have been proposed to fit complex data better, the advances of science and technology have generated more complex data, e.g., Big Data, in which existing probability theory and statistical models find their limitations. This work establishes probability foundations for studying extreme values of data generated from a mixture process with the mixture pattern depending on the sample length and data generating sources. In particular, we show that the limit distribution, termed as the accelerated max-stable distribution, of the maxima of maxima of sequences of random variables with the above mixture pattern is a product of three types of extreme value distributions. As a result, our theoretical results are more general than the classical extreme value theory and can be applicable to research problems related to Big Data. Examples are provided to give intuitions of the new distribution family. We also establish mixing conditions for a sequence of random variables to have the limit distributions. The results for the associated independent sequence and the maxima over arbitrary intervals are also developed. We use simulations to demonstrate the advantages of our newly established maxima of maxima extreme value theory.

• Beirlant, J., Goegebeur, Y., Segers, J., & Teugels, J. (2004). Statistics of extremes: Theory and applications. Wiley Series in Probability and Statistics. Wiley. [Crossref], [Google Scholar]
• Bücher, A., & Segers, J. (2017). On the maximum likelihood estimator for the generalized extreme-value distribution. Extremes, 20(4), 839–872. https://doi.org/10.1007/s10687-017-0292-6 [Crossref], [Web of Science ®], [Google Scholar]
• Castillo, E., Hadi, A. S., Balakrishnan, N., & Sarabia, J. M. (2005). Extreme value and related models with applications in engineering and science. Wiley Series in Probability and Statistics. Wiley. [Google Scholar]
• Chavez-Demoulin, V., Embrechts, P., & Hofert, M. (2016). An extreme value approach for modeling operational risk losses depending on covariates. Journal of Risk and Insurance, 83(3), 735–776. https://doi.org/10.1111/jori.v83.3 [Crossref], [Web of Science ®], [Google Scholar]
• Coles, S. (2001). An introduction to statistical modeling of extreme values. Springer. [Crossref], [Google Scholar]
• Daouia, A., Girard, S., & Stupfler, G. (2018). Estimation of tail risk based on extreme expectiles. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 80(2), 263–292. https://doi.org/10.1111/rssb.12254 [Crossref], [Web of Science ®], [Google Scholar]
• de Haan, L. (1993). Extreme value statistics. In J. Galambos, J. Lechner, & E. Simiu (Eds.), Extreme value theory and applications (pp. 93–122). Kluwer Academic Publisher. [Google Scholar]
• de Haan, L., & Ferreira, A. (2006). Extreme value theory: An introduction. Springer Verlag. [Crossref], [Google Scholar]
• Dey, D. K., & Yan, J. (2016). Extreme value modeling and risk analysis: Methods and applications. Chapman & Hall/CRC. [Crossref], [Google Scholar]
• Embrechts, P., Resnick, S. I., & Samorodnitsky, G. (1999). Extreme value theory as a risk management tool. North American Actuarial Journal, 3(2), 30–41. https://doi.org/10.1080/10920277.1999.10595797 [Taylor & Francis Online], [Google Scholar]
• Finkenstädt, B., & Rootzén, H. E. (2004). Extreme values in finance, telecommunications, and the environment. Chapman & Hall/CRC. [Google Scholar]
• Galambos, J. (1978). The asymptotic theory of extreme order statistics. Technical Report. [Google Scholar]
• Heffernan, J. E., Tawn, J. A., & Zhang, Z. (2007). Asymptotically (in)dependent multivariate maxima of moving maxima processes. Extremes, 10(1–2), 57–82. https://doi.org/10.1007/s10687-007-0035-1 [Crossref], [Google Scholar]
• Idowu, T., & Zhang, Z. (2017). An extended sparse max-linear moving model with application to high-frequency financial data. Statistical Theory and Related Fields, 1(1), 92–111. https://doi.org/10.1080/24754269.2017.1346852 [Taylor & Francis Online], [Google Scholar]
• Leadbetter, M. R., Lindgren, G., & Rootzén, H. (2012). Extremes and related properties of random sequences and processes. Springer Science & Business Media. [Google Scholar]
• Loynes, R. M. (1965). Extreme values in uniformly mixing stationary stochastic processes. The Annals of Mathematical Statistics, 36(3), 993–999. https://doi.org/10.1214/aoms/1177700071 [Crossref], [Google Scholar]
• Malinowski, A., Schlather, M., & Zhang, Z. (2015). Marked point process adjusted tail dependence analysis for high-frequency financial data. Statistics and Its Interface, 8(1), 109–122. https://doi.org/10.4310/SII.2015.v8.n1.a10 [Crossref], [Web of Science ®], [Google Scholar]
• Mao, G., & Zhang, Z. (2018). Stochastic tail index model for high frequency financial data with Bayesian analysis. Journal of Econometrics, 205(2), 470–487. https://doi.org/10.1016/j.jeconom.2018.03.019 [Crossref], [Web of Science ®], [Google Scholar]
• McNeil, A. J., & Frey, R. (2000). Estimation of tail-related risk measures for heteroscedastic financial time series: An extreme value approach. Journal of Empirical Finance, 7(3–4), 271–300. https://doi.org/10.1016/S0927-5398(00)00012-8 [Crossref], [Google Scholar]
• Mikosch, T., Embrechts, P., & Klüppelberg, C. (1997). Modelling extremal events for insurance and finance. Springer Verlag. [Google Scholar]
• Naveau, P., Zhang, Z., & Zhu, B. (2011). An extension of max autoregressive models. Statistics and Its Interface, 4(2), 253–266. https://doi.org/10.4310/SII.2011.v4.n2.a19 [Crossref], [Web of Science ®], [Google Scholar]
• Resnick, S. I. (2013). Extreme values, regular variation and point processes. Springer. [Google Scholar]
• Rocco, M. (2014). Extreme value theory in finance: A survey. Journal of Economic Surveys, 28(1), 82–108. https://doi.org/10.1111/joes.2014.28.issue-1 [Crossref], [Web of Science ®], [Google Scholar]
• Rosenblatt, M. (1956). A central limit theorem and a strong mixing condition. Proceedings of the National Academy of Sciences USA, 42(1), 43–47. https://doi.org/10.1073/pnas.42.1.43 [Crossref], [Web of Science ®], [Google Scholar]
• Salvadori, G., De Michele, C., Kottegoda, N. T., & Rosso, R. (2007). Extremes in nature: An approach using copulas. Springer(Complexity). [Crossref], [Google Scholar]
• Smith, R. L. (1985). Maximum likelihood estimation in a class of nonregular cases. Biometrika, 72(1), 67–90. https://doi.org/10.1093/biomet/72.1.67 [Crossref], [Web of Science ®], [Google Scholar]
• Tang, R., Shao, J., & Zhang, Z. (2013). Sparse moving maxima models for tail dependence in multivariate financial time series. Journal of Statistical Planning and Inference, 143(5), 882–895. https://doi.org/10.1016/j.jspi.2012.11.008 [Crossref], [Web of Science ®], [Google Scholar]
• Tsay, R. S. (2005). Analysis of financial time series (Vol. 543). John Wiley & Sons. [Crossref], [Google Scholar]
• Watson, G. S. (1954). Extreme values in samples from m-dependent stationary stochastic processes. The Annals of Mathematical Statistics, 25(4), 798–800. https://doi.org/10.1214/aoms/1177728670 [Crossref], [Google Scholar]
• Zhang, Z., & Smith, R. L. (2010). On the estimation and application of max-stable processes. Journal of Statistical Planning and Inference, 140(5), 1135–1153. https://doi.org/10.1016/j.jspi.2009.10.014 [Crossref], [Web of Science ®], [Google Scholar]
• Zhang, Z., & Zhu, B. (2016). Copula structured m4 processes with application to high-frequency financial data. Journal of Econometrics, 194(2), 231–241. https://doi.org/10.1016/j.jeconom.2016.05.004 [Crossref], [Web of Science ®], [Google Scholar]
• Zhao, Z., & Zhang, Z. (2018). Semi-parametric dynamic max-copula model for multivariate time series. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 80(2), 409–432. https://doi.org/10.1111/rssb.12256 [Crossref], [Web of Science ®], [Google Scholar]
• Zhao, Z., Zhang, Z., & Chen, R. (2018). Modeling maxima with autoregressive conditional Fréchet model. Journal of Econometrics, 207(2), 325–351. https://doi.org/10.1016/j.jeconom.2018.07.004 [Crossref], [Web of Science ®], [Google Scholar]

​​​​​​To cite this article: Wenzhi Cao & Zhengjun Zhang (2021) New extreme value theory for maxima of maxima, Statistical Theory and Related Fields, 5:3, 232-252, DOI:10.1080/24754269.2020.1846115
To link to this article: https://doi.org/10.1080/24754269.2020.1846115