UDC 303.72: 005.7
DOI: 10.36871/2618-9976.2023.03.005

Authors

Ilya M. Tolstobrov,
2nd year student of the Department of Data Analysis and Machine Learning of the Faculty of Information Technology and Big Data Analysis of the Financial University under the Government of the Russian Federation, majoring in Applied Mathematics and Informatics
Yulia B. Kamalova,
Senior Lecturer, Department of Data Analysis and Machine Learning, Faculty of Information Technology and Big Data Analysis, Financial University under the Government of the Russian Federation

Abstract

In the work, the movement of the stock market in different countries during the crisis year of 2020 was studied, which did not depend on the decisions made in this country, but obeyed the general trend. In particular, the behavior of sectoral indices of different countries during the first year of the coronavirus was studied and it was concluded that the movement of markets is subject to the global trend.
For the analysis, sectoral indices of countries affected to varying degrees by the coronavirus were selected, whose securities markets received various support measures: the United States, Russia, China, Japan and Germany. The impact of the pandemic on industries such as financial services, metallurgy, chemicals, consumer goods and telecommunications is considered.
Quotes are considered for the period from 01.01.2020/31.12.2020/XNUMX to XNUMX/XNUMX/XNUMX. Graphs of the values of sectoral indices of the Russian Federation and the USA in the specified period are presented and analyzed. Graphs are built in the Jupyter Notebook environment using the Python language and the plotly data visualization library. The concept of hypothesis and criterion is defined and the most important definitions are given. With a decrease in the probability of making a type XNUMX error, the probability of making a type XNUMX error increases, so the significance level is chosen so that the power of the criterion is maximum.
To test the hypothesis, we used the multivariate KruskalWallis test and its modified version for mediumsized samples with a significance level of α=0,05. The novelty of the study lies in the application of the KruskalWallis criterion to financial data. The study was conducted in Python in the Jupyter Notebook environment using the numpy, pandas, itertools, scipy and plotly libraries.

Keywords

Kruskal-Wallis test, Stock markets, Python, Numpy, Pandas, Itertools, Scipy and plotly