Statistical Research of Esports Industry

The research provides quantitative analysis of new computerized industry called eSports. Compositionally research consists of two parts divided with respect to scale. The first part highlights the usage of quantitative information and statistical methods for estimating the size of the industry and making forecast. The second part closely interconnected with concrete Esports discipline- Dota 2. One of the main objectives of the first part of the study is to show that the electronic sports industry has been already a complete economic market with lots of institutes and individual agents, such as: gaming leagues, teams and eSports organizations, professional players, coaches, commentators etc. Then after aggregating values, which quantitively describes the contemporary size of eSports market, the time series models are used for making a forecast in a short run. To define some structural changes from 2015 to 2016 a model of mixture of distributions is used. The second part of the research provides econometric analysis and usage of binary choice regressions to predict the outcomes of professional games in Dota 2 among professional teams using their inner quantitative indicators as factors in prediction model. The prediction model was estimated with the usage of sample of more than 1500 professional games. As a result, a Gompit model with sufficiently high predictive ability was chosen as optimal, which helps to forecast the outcomes for Dota 2 professional games. Moreover, time series models are also applied for forecasting the audience of ordinary players for the chosen discipline.