Multifractal analysis of social media use in financial markets

We analyze the nonlinear properties of social media activity(SMA) using the multifractal detrended fluctuation analysis (MF-DFA) method. Social media data related to the stock market are gathered from social media platforms. Using data on over 2000 firms in the Korean stock market for 2018-2020, we study social media activity and its differences to evaluate associated nonlinear and statistical properties. We find that the cumulative distribution function of SMA follows a stretched exponential distribution with $$ . The Hurst exponent of SMA for three datasets (2018, 2019, 2020 year) is larger than 0.9, whereas the Hurst exponents of shuffled time series have values of approximately 0.5. In particular, we find a multifractal structure in both SMA and SMA difference results irrespective of the period and degree of multifractality defined as $$ , which reaches a maximum value during the COVID-19 pandemic as a financial crisis.