Statistical properties of online avatar numbers in a massive multiplayer online role-playing game

TL;DR

This paper analyzes the statistical properties of online avatar numbers in an MMORPG, revealing daily patterns, non-Gaussian fluctuations, long-term correlations, and multifractal behavior, which are crucial for understanding game popularity dynamics.

Contribution

It provides a comprehensive statistical analysis of online avatar numbers, uncovering their periodic, non-Gaussian, and multifractal characteristics, which were not previously documented.

Findings

Online avatar numbers show daily periodicity and intraday patterns.

Fluctuations follow a leptokurtic, power-law tail distribution.

Time series exhibit long-term correlations and multifractality.

Abstract

Massive multiplayer online role-playing games (MMORPGs) are very popular in past few years. The profit of an MMORPG company is proportional to how many users registered, and the instant number of online avatars is a key factor to assess how popular an MMORPG is. We use the on-off-line logs on an MMORPG server to reconstruct the instant number of online avatars per second and investigate its statistical properties. We find that the online avatar number exhibits one-day periodic behavior and clear intraday pattern, the fluctuation distribution of the online avatar numbers has a leptokurtic non-Gaussian shape with power-law tails, and the increments of online avatar numbers after removing the intraday pattern are uncorrelated and the associated absolute values have long-term correlation. In addition, both time series exhibit multifractal nature.