Registration

* - required fields

Password changing

The new password
was sent

We have sent you an email with a password

Joint Seminar of the Department of Probability Theory & Vega Foundation

The scientific and educational seminar is held jointly by the Department of Probability Theory of Lomonosov Moscow State University and the Vega Institute Foundation on a regular basis on Wednesdays. 

The speakers of the seminar introduce the listeners to the latest achievements in the field of financial and actuarial mathematics.

The seminar is primarily aimed at undergraduate students of last two years, graduate students and PhD students. However, everyone can take part, having passed the preliminary registration.

The leaders of the seminar are Academician of the Russian Academy of Sciences Albert Nikolaevich Shiryaev, Chief Executive Officer of the Foundation Kirill Yuryevich Klimov and Senior Researcher of the Steklov Mathematical Institute of Russian Academy of Sciences Mikhail Valentinovich Zhitlukhin

The schedule is subject to change

Language: Russian, English
Format: online

SCHEDULE FOR THE SPRING SEMESTER' 23 

May 17, 18:30-20:00 

 Rustam IBRAGIMOV 
Imperial College Business School and the Centre for Econometrics and Business Analytics

New Approaches to Robust Inference on Market (Non-)Efficiency, Volatility Clustering and Nonlinear Dependence

We present novel, robust methods for inference on market (non-)efficiency, volatility clustering, and nonlinear dependence in financial return series. In contrast to existing methodology, our proposed methods are robust against non-linear dynamics and tail-heaviness of returns. Specifically, our methods only rely on return processes being stationary and weakly dependent (mixing) with finite moments of a suitable order. This includes robustness against power law distributions associated with non-linear dynamic models such as GARCH and stochastic volatility. The methods are easy to implement and perform well in realistic settings. We revisit a recent study by Baltussen et al. (2019, Journal of Financial Economics, vol. 132, pp. 26-48) on autocorrelation in major stock indexes. Using our robust methods, we document that the evidence of presence of negative autocorrelation is weaker, compared to the conclusions of the original study.

WATCH IT IN THE RECORDING


May 10, 18:30-20:00 

 Mikhail ZHITLUKHIN 
Candidate of Physical and Mathematical Sciences, Senior Researcher, Steklov Mathematical Institute of Russian Academy of Sciences

Evolutionary optimal strategies in dynamic stochastic games

In the presentation one class of games arising in mathematical economics will be considered. We will introduce the notion of the evolutionary optimal strategy as such strategy that the proportion of players using it remains separated from zero with probability 1 on the infinite time horizon regardless of which strategies the other players use. The main results are proof of the existence of such strategies for the considered game class and evaluation of the proximity in some sense of strategies satisfying this property.

WATCH IT IN THE RECORDING


May 3, 18:30-20:00 

 Rostislav BEREZOVSKY 
Hash CIB

Exchange models in decentralized finance

The presentation addresses the examples of basic decentralized financial services and the general form of the CFMM-like decentralized exchange. The exchange with concentrated liquidity and the related problems of liquidity provider are considered in detail as well as the promising new instruments arising as a combination of the existing models.

WATCH IT IN THE RECORDING


April 19, 18:30-20:00 

 Victor ANTIPOV  
Scientific supervisor: Yury Mikhailovich Kabanov

On the probability of ruin with investments in a risky asset

 Platon PROMYSLOV  
Scientific supervisor: Yury Mikhailovich Kabanov

Ruin probability in Sparre Andersen model with investments: annuity payments case
 
 Alexandra TOKAEVA 
Scientific supervisor: Mikhail Valentinovich ZHITLUKHIN

Optimal growth strategy in the multiagent market model with affine payouts 

 Alexey SHATOHIN  
Scientific supervisor: Ekaterina Vadimovna Bulinskaya

Estimate of the insurer and reinsurer ruin probability in the risk model with quota share reinsurance

WATCH IT IN THE RECORDING


April 12, 18:30-20:00 

 Dmitry ROKHLIN 
Professor of the Department of Higher Mathematics and Operations Research, Southern Federal University

Online optimization and some applications

Methods of convex online optimization allows for construction of theoretically grounded recurrent algorithms for solving optimization problems involving unknown varying factors. The principles for constructing such algorithms will be discussed as well as a number of applications: dynamic portfolio problem, the problem of finding incentive prices and regression problem with random features.

WATCH IT IN THE RECORDING


April 5, 18:30-20:00 

 Ivan POLOZOV  
Scientific supervisor: Valentin Dmitrievich Konakov

Markov chains related to Robbins–Monro procedure

 Ruslan GURIEV  
Scientific supervisor: Ekaterina Mikhailovna Ryadnova

Extrema of random networks and their connection with p2p-exchanges
 
 Valentin KUZMENKO 
Scientific supervisor: Ekaterina Mikhailovna Ryadnova

Analysis of financial data with an index of extreme values close to 0 

 Gordey VERBIY  
Scientific supervisor: Maksim Petrovich Zapletin

Stochastic optimization of portfolio investments 

WATCH IT IN THE RECORDING


March 29, 18:30-20:00 

 Alexander RAKITKO 
Graduate of the Faculty of Mechanics and Mathematics of Lomonosov Moscow State University, medical genetic center Genotek
Scientific supervisor: A. Bulinsky


Identification of significant collections of factors with error functional estimation

The problem of multifactorial dimensionality reduction (MDR) arises naturally in different applications. We consider MDR-EFE (error function estimation) method for the identification of significant collections of factors. The asymptotic results (criterion of almost sure convergence, several variants of central limit theorems and others) are proven using the martingale technique, theory of exchangeable variables for various modifications of MDR-EFE method. Theoretical results are supported by computer simulations.

WATCH IT IN THE RECORDING


March 22, 18:30-20:00 

 Makar PRIKHODKO 
Scientific supervisor: Ekaterina Vadimovna Bulinskaya

Analysis of methods for calculating insurance premiums

 Andrey KANAEV 
Scientific supervisor: Elena Borisovna Yarovaya

Application of branching processes in financial mathematics

 Danila SHABALIN 
Scientific supervisor: Ekaterina Vadimovna Bulinskaya

Research of ruin probability for some insurance models


WATCH IT IN THE RECORDING


March 15, 18:30-20:00 

 Askhab SULEYMANOV 
Scientific supervisor: Elena Evgenievna Bashtova

The probability of non-ruin by the t-time in the insurance model with capital investment 

 Anastasia  MAKHOVA 
Scientific supervisor: Mikhail Ivanovich Kumskov

Forecasting financial time-series using neural networks with memory

 Vladislav ZYUZIN 
Scientific supervisor: Mikhail Ivanovich Kumskov

Forecasting financial time-series

WATCH IT IN THE RECORDING


March 01, 18:30-20:00 

 Vsevolod ZAOSTROVSKII  
Scientific supervisor: Yury Mikhailovich Kabanov

Local and stochastic volatility models

 Ivan CHEREPAKHIN  
Scientific supervisor: Yury Mikhailovich Kabanov

Breakdown of the Wiener process using wavelet

 Diana KALIKAEVNA
Scientific supervisor: Alexander Yurievich Veretennikov

Recursive properties of "Markov-up" processes

WATCH IT IN THE RECORDING


SCHEDULE FOR THE FALL SEMESTER'22

October 19, 18:10-19:30

 Albert Nikolaevich SHIRYAEV  ,
Academician of the Russian Academy of Sciences, Doctor of Physical and Mathematical Sciences, Head of the Department of Probability Theory

Randomness in probability

WATCH IT IN THE RECORDING


October 26, 18:10-19:30

 Ivan Valerievich OSELEDETS ,
Doctor of Physical and Mathematical Sciences, Professor of the Russian Academy of Sciences, Director of the Skoltech AI Technology Center, Leading Researcher, Institute of Computational Mathematics of the Russian Academy of Sciences and AIRI 
 
Methods of tensor expansions and their applications

The lecture will give an overview of basic approaches for efficient representations of multidimensional arrays and some applications of such approaches.

WATCH IT IN THE RECORDING


November 02, 18:10-19:30

 Vladimir Vladimirovich PITERBARG ,
PhD, Global Head Of Quantitative Analytics, NatWest Markets

Alternatives to Deep Neural Networks for Function Approximations in Finance: Function Fitting and Regressions

We develop two methods for approximating slow-to-calculate functions, and for conditional expected value calculations: the generalized stochastic sampling (gSS) and the functional tensor train (fTT) methods. We propose them as highly-performing alternatives to generic deep neural networks (DNNs) currently routinely recommended in derivatives pricing and other quantitative finance applications. The new methods not only outperform DNNs for typical financial problems but also, unlike DNNs, satisfy stringent finance requirements such as predictability and explainability.

WATCH IT IN THE RECORDING


November 09, 18:10-19:30

 Platon Valerievich  PROMYSLOV ,
Doctoral student of the Department of Probability Theory, Faculty of Mechanics and Mathematics, Moscow State University

Sparre Andersen's Bankruptcy Models with Investments

WATCH IT IN THE RECORDING


November 16, 18:10-19:30

 Anna Alexandrovna OBIZHAEVA ,
Ph.D., MIT Sloan, Professor, Director of the NES Master of Finance program

Dimensional Analysis, Leverage Neutrality, and Market Microstructure Invariance

This paper combines dimensional analysis, leverage neutrality, and a principle of market microstructure invariance to derive scaling laws expressing bid-ask spreads, transaction costs functions, bet sizes, number of bets, and other financial variables in terms of dollar trading volume and volatility. The scaling laws are tested using data on bid-ask spreads for Russian and U.S. stocks. These scaling laws provide practical metrics for risk managers and traders; scientific benchmarks for evaluating controversial issues related to high frequency trading, market crashes, and liquidity measurement; and guidelines for designing policies in the aftermath of financial crisis.

WATCH IT IN THE RECORDING


November 23, 18:10-19:30

 Andrey Lvovich ITKIN ,
Doctor of Physical and Mathematical Sciences,Professor of NYC, USA; Quantitative research and development Lead, ADIA, UAE

The ATM implied skew in the ADO-Heston model

Rough volatility (RV) models increase their popularity since 2007 when it was first shown that for a wide range of assets, historical volatility time-series exhibit a behavior which is much rougher than that of a Brownian motion (BM). One of the important findings of the RV models consists in their ability to reproduce the explosive behavior of the implied at-the-money skew observed empirically when the option maturity goes to zero. On the other hand, the cost one has to pay for getting the advantages of a RV model are technical problems arising due to the non-Markovian nature of the fractional BM. 

At the same time, alternative to RV and simpler models have been proposed in the literature, which are Markovian in nature, and, thus, allow solving pricing problems via a well-elaborated approaches, e.g. by solving a PDE. Moreover, recent analysis of market data and numerical experiments show that, even when the instantaneous volatility has diffusive dynamics with the same roughness as the BM, the realized volatility exhibits rough behavior corresponding to a Hurst exponent significantly smaller than 0.5. Also, it is reported that the implied ATM skew does not follow a power law for short maturities and is better captured by simple parametrizations that do not blow up for vanishing maturity.

From the modeling point of view these results mean that, perhaps, market data on realized volatility cannot be used to decide which one - rough or Markovian stochastic volatility model is preferable to replicate the observed market behavior. Therefore, other measures would be useful for this purpose, e.g. the vanilla and forward implied volatilities and skews which could be retrieved from the market data. 

In this presentation we discuss the above problems in more detail, and also, following the idea of [P. Carr, A. Itkin, 2019, Risk], describe a new Markovian approximation of the rough Heston model. We show that the behavior of the implied ATM skew in this model is similar to that reported in [M. Amrani, J. Gyon, 2022] with no  blow up for vanishing maturity. 

WATCH IT IN THE RECORDING


November 30, 18:10-19:30

  Evgeny Vladimirovich BURNAEV ,
Doctor of Physical and Mathematical Sciences, Professor, Head of the Center of Applied AI Skoltech, Head of the scientific group, AIRI

Topology strikes back: everything you wanted to know about the shape of your data but were afraid to ask

Real world data has a shape, and a shape "makes a difference." However, standard machine learning methods often do not take into account the shape of the data. In turn, modern methods of topological data analysis just analyze their form as the most important property. The lecture will discuss how topological data analysis works. It will be shown how topological features make it possible to describe the shape of data and significantly increase the efficiency of machine learning models for processing information from seemingly completely different areas - text data and speech data. Based on a new topological measure of similarity, an approach will be described for constructing a low-dimensional description of a data shape that has the property of "disentanglement" - various parameters of this description are automatically responsible for different properties of the data, which makes it possible to increase the interpretability of machine learning models.

WATCH IT IN THE RECORDING


December 07, 18:10-19:30

 Viktor ANTIPOV ,
Doctoral student of the Department of Probability Theory, Faculty of Mechanics and Mathematics, Moscow State University

On ruin probabilities with investments in a risky asset

In this paper, we study the asymptotics of the ruin probability in the Cramer — Lundberg model with exponentially distributed payments and capital investment in a risky asset. It is shown how, using the methods of power geometry, one can obtain power asymptotics for solutions of differential equations describing the ruin probability as a function of the initial capital. For the case when the dynamics of an asset is given by a geometric Brownian motion, the corresponding boundary value problems are set and formulas for asymptotic expansions for the ruin probability are obtained. In the case when the parameters of a geometric Brownian motion obey a two-state Markov process, an equation is obtained for the ruin probability asymptotic index.

WATCH IT IN THE RECORDING


December 14, 18:10-19:30

 Artur SIDORENKO ,
Doctoral student of the Department of Probability Theory, Faculty of Mechanics and Mathematics, Moscow State University

Axiomatic view of the Rogers–Veraart and Suzuki–Elsinger models of systemic risk

An interbank network model with cross-holdings and default costs is studied. Following the approach of Eisenberg and Noe, we define the model using a few natural financial rules and derive a finite family of fixed point problems. These fixed-point problems are parameterized by Boolean cube vectors, which can be interpreted as the propensity of banks to fictitious bankruptcy. The proposed model combines the main properties of the Rogers–Veraart, Suzuki–Elsinger, and Ararat–Meimandzhanov models. Methods are proposed for calculating the maximum and minimum clearing pairs using mixed-integer linear programming and the Gaussian algorithm for eliminating variables.

WATCH IT IN THE RECORDING


You can watch the meetings that took place earlier in the playlist of the Foundation's channel

BACK TO THE COURSE LIST