2026 CRM-SSC Prize awarded to Stanislav Volgushev (University of Toronto)

Professor Volgushev was born in Moscow in 1983 and moved with his parents to Germany at the age of 6. He studied Mathematics at the Ruhr University Bochum and received his diplom (equivalent of MSc) in 2007 and his PhD in 2010, both under the supervision of Holger Dette. He joined Cornell University as Assistant Professor in 2015 and moved to Toronto in 2016.

Professor Volgushev has a total of 48 publications, many appearing in the leading journals of the discipline,  including the Annals of Statistics (AoS) , the Journal of the Royal Statistical Society (JRSSB), and the Journal of the American Statistical Association (JASA).

In his PhD, he worked on quantile regression, and he has continued this line of research throughout his career. Among other contributions on this theme, in his widely cited 2019 AoS paper (joint with Guang Cheng and Shi-Kang Chao), Volgushev proposed the first approach to quantile regression for very large data sets utilizing the divide-and-conquer approach and not only established guarantees for the success of divide and conquer procedures but also exhibited scenarios where divide and conquer procedures provably fail. He also made significant contributions to quantile regression for panel data with Jiaying Gu. In his 2020 Journal of Econometrics publication with Jiaying Gu and Antonio Galvao, he established new results on the limiting distribution of quantile regression when many individual specific effects are estimated simultaneously. He has also worked on several other aspects of dependence modeling. His work on Hadamard differentiability of the copula map (Bücher, Volgushev, JMVA, 2013), convergence in weak metrics under mild smoothness assumptions (Bücher, Segers, Volgushev, AoS 2014), and weak convergence with respect to stronger weighted metrics (Berghaus, Bücher, Volgushev, Bernoulli 2017) provide key tools for the analysis of rank-based procedures. Another line of his work, joint with Marc Hain, Holger Dette and Tobias Key, combines the power of copulas with frequency domain methods in time series.  Many of Professor Volgushev’s recent contributions are to extreme value analysis approaching several questions including the dichotomy between asymptotic independence and dependence (Lalancette, Engelke, Volgushev, AoS 2021), tree structure learning (Engelke, Volgushev, JRSSB 2022) and learning flexible graphical models for extremes (Engelke, Lalancette, Volgushev, AoS 2026).

Volgushev greatly enjoys working and learning together with trainees. During his time at the University of Toronto four PhD students have completed their degrees under his (co-)supervision and he is currently supervising four PhD students and one Postdoc. He has contributed to the Department of Statistical Sciences by serving as associate chair for graduate studies and to the wider profession as associate editor for several journals, including Bernoulli, the Electronic Journal of Statistics, the Canadian Journal of Statistics and Extremes.