?
Application of Kalman Filter with alpha-stable distibution
P. 419–427.
Mozgunov P.
In this paper we consider the behavior of Kalman Filter state estimates in the case of distribution with heavy tails .The simulated linear state space models with Gaussian measurement noises were used. Gaussian noises in state equation are replaced by components with alpha-stable distribution with different parameters alpha and beta. We consider the case when "all parameters are known" and two methods of parameters estimation are compared: the maximum likelihood estimator (MLE) and the expectation- maximization algorithm (EM). It was shown that in cases of large deviation from Gaussian distribution the total error of states estimation rises dramatically. We conjecture that it can be explained by underestimation of the state equation noises covariance matrix that can be taken into account through the EM parameters estimation and ignored in the case of ML estimation.
Keywords: EM-алгоритмEM-algorithmKalman filterфильтр КалманаAlpha-stable distibutionHeavy-tailed distibutionsАльфа-устойчивое распределениеРаспределение с тяжелыми хвостами
Publication based on the results of:
Eliseev A., В кн.: Системное моделирование социально-экономических процессов: труды 45-ой Юбилейной международной научной школы-семинара, д. Красновидово Московской области, 3 – 9 октября 2022 г.: Воронеж: Истоки, 2022. С. 400–403.
Макроэкономические прогнозирование на основе моделей общего равновесия имеет ряд неоспоримых преимуществ. Однако из-за использования на квартальных данных данные модели не позволяют учитывать более оперативную информацию из более высокочастотных данных (например, ежемесячных), тогда как ранняя идентификация происходящих в экономике шоков является важным условием выстраивания рекомендаций для макроэкономической политики. Настоящее исследование решает данную проблему, используя методологию комбинирования ...
Added: June 9, 2023
Deryugina E., Maria Guseva, Alexey Ponomarenko, Journal of Central Banking Theory and Practice 2022 Vol. 11 No. 1 P. 87–104
We conduct a Monte Carlo experiment using an ad-hoc New Keynesian model and a tractable agent-based model to generate artificial credit cycle episodes. We show that fluctuations in the implicit measures of the natural rate of interest obtained using a conventional trivariate Kalman filter on these artificial datasets occur in the vicinity of credit cycle ...
Added: January 31, 2023
Zinchenko D., Zinchenko A., E. Nikonov, Physics of Particles and Nuclei 2022 Vol. 53 No. 2 P. 519 –523
Although the start-up of the MPD experiment is still ahead, the work on the preparation and physics validation of the future detector upgrade program has already been initiated. As one of the possible MPD upgrade steps, an Inner Tracking System (ITS) based on the next generation silicon pixel detectors is being considered to be installed ...
Added: January 12, 2023
Marino L., Menozzi S., Priola E., Studia Mathematica 2022 Vol. 267 No. 3 P. 321–346
We consider a possibly degenerate Kolmogorov Ornstein–Uhlenbeck operator of the form L=Tr(BD2)+⟨Az,D⟩, where A, B are N×N matrices, z∈RN, N≥1, which satisfy the Kalman condition which is equivalent to the hypoellipticity condition. We prove the following stability result: the Schauder and Sobolev estimates associated with the corresponding parabolic Cauchy problem remain valid, with the same constant, for the parabolic Cauchy problem associated with ...
Added: November 10, 2022
Gorbunov E., Danilova M., Shibaev I. et al., / Series arXiv:2106.05958 "arXiv:2106.05958". 2021.
Thanks to their practical efficiency and random nature of the data, stochastic first-order methods are standard for training large-scale machine learning models. Random behavior may cause a particular run of an algorithm to result in a highly suboptimal objective value, whereas theoretical guarantees are usually proved for the expectation of the objective value. Thus, it ...
Added: October 25, 2021
Moiseev N., Sorokin A., Zvezdina N. et al., Mathematics 2021 No. 9(19) Article 2423
The research paper is devoted to developing a mathematical approach for dealing with time-varying parameters in rolling window logit models for credit risk assessment. Forecasting coefficients yields a better model accuracy than a trivial approach of using computed past statistics parameters for the next time period. In this paper, a new method of dealing with ...
Added: October 1, 2021
Pogorelova P., Peresetsky A., Прикладная эконометрика 2020 Т. 57 С. 53–71
In this paper, the Kalman linear filter method is used to decompose non‐synchronous observations of the realized volatility of financial indices (NIKKEI 225, FTSE 100, S&P 500) into unobservable global and local components. It is shown that the volatility of the New York S&P 500 index is a global component, while the Tokyo NIKKEI 225 ...
Added: August 26, 2020
Molchanov S., Panov V., Stochastics-An International Journal of Probability and Stochastic Processes 2019 Vol. 91 No. 5 P. 754–772
In this paper, we consider limit laws for the model, which is a generalisation of the random energy model (REM) to the case when the energy levels have the mixture distribution. More precisely, the distribution of the energy levels is assumed to be a mixture of two normal distributions, one of which is standard normal, ...
Added: November 19, 2018
Ivlev N., Ovchinnikov M., Ivanov D. et al., Acta Astronautica 2014 Vol. 93 P. 23–33
The key-problems of design, examination, laboratory' and flight testing of the attitude determination and control system (ADCS) dedicated for a microsatellite arc considered. The system consists of three pairs of the reaction wheels, three magnetorquers, set of Sun sensors, three-axis magnetometer and a control unit. ADCS, on one hand, is subjected to the high accuracy ...
Added: June 8, 2018
Peresetsky A., Turmuhambetova G., Urga G., Emerging Markets Review 2001 Vol. 2 No. 1 P. 1–16
This study analyzes two issues related to the GKO futures market in Russia in 1996 and 1997. First, we evaluate the existence of a risk premium in this market. We show its existence providing a functional form for the premium. The main result is that risk premium depends positively on the time before delivery of ...
Added: April 16, 2018
Akimov P. A., Matasov A. I., IEEE Transactions on Automatic Control 2015 Vol. 60 No. 4 P. 1050–1063
The mixed-norm cost functions arise in many applied optimization problems. As an important example, we consider the state estimation problem for a linear dynamic system under a nonclassical assumption that some entries of state vector admit jumps in their trajectories. The estimation problem is solved by means of mixed l1/l2-norm approximation. This approach combines the ...
Added: November 5, 2017
Asaturov K. G., Journal of Applied Economic Sciences 2016 Vol. XI No. 8(46) P. 1624–1642
This paper examines the dynamic beta of Russian companies within the framework of the market model. The closing weekly prices of 29 Russian stocks, six Russian sector indices and the MICEX Index as a market index during the period from January 2009 to June 2015 are used to estimate time-varying beta using various econometric techniques. ...
Added: March 22, 2017
Peresetsky A., Yakubov R., International Journal of Computational Economics and Econometrics 2017 Vol. 7 No. 1-2 P. 152–169
In this paper, a Kalman filter-type model is used to extract a global stochastic trend from discrete non-synchronous data on daily stock market index returns from different markets. The model allows for the autocorrelation in the global stochastic trend, which means that its increments are predictable. It does not necessarily mean the predictability of market ...
Added: January 4, 2017
Veretennikova M., Kolokoltsov V., Journal of Applied Nonlinear Dynamics 2017 Vol. 6 No. 1 P. 45–56
In this paper we initiate the rigorous analysis of controlled Continuous Time Random Walks (CTRWs) and their scaling limits, which paves the way to the real application of the research on CTRWs, anomalous diffusion and related processes. For the first time convergence is proven for payoff functions for controlled scaled CTRWs and their position dependent ...
Added: October 5, 2016
Veretennikova M., Kolokoltsov V., Communications in Applied and Industrial Mathematics 2015 Vol. 6 No. 1 P. 1–18
In this article we study a controlled Continuous Time Random Walk and its position-dependent extensions. We heuristically derive the optimal payoff function equations for their scaling limits. The general equation for the corresponding optimal payoff of the limiting process may be called a fractional Hamilton Jacobi Bellman equation. ...
Added: October 5, 2016
Alexander Yu. Apokin, Irina B. Ipatova, / Series WP BRP "Basic research program". 2016. No. 142.
This paper investigates the factors behind the recent growth slowdown (so-called Secular Stagnation) in the US, the euro area and Japan using the metrics of potential output growth. Specifically, our results offer limited support for an impaired credit transmission channel hypothesis (Reinhart and Rogoff, 2009a), while not supporting a supply slowdown hypothesis (Gordon, 2012). We ...
Added: September 9, 2016
Andresen A., Spokoiny V., Journal of Machine Learning Research 2016 No. 17(63) P. 1–53
We derive two convergence results for a sequential alternating maximization procedure to approximate the maximizer of random functionals such as the realized log likelihood in MLE estimation. We manage to show that the sequence attains the same deviation properties as shown for the profile M-estimator by Andresen and Spokoiny (2013), that means a finite sample ...
Added: September 8, 2016
Asaturov K. G., Экономика и математические методы 2015 Т. 51 № 4 С. 59–75
The paper examines dynamic systematic risk nature of Indian companies in the frame of the market model. The closing weekly prices of 89 Indian stocks and BSE 100 Index as the market index during the period from January 2000 to December 2013 are analyzed with rolling OLS, multivariate GARCH models, semiparametric regression and a Kalman ...
Added: December 5, 2015