Economists Suggest Using Media's Attention to Bitcoin to Predict its Returns

Researchers at the HSE Faculty of Economic Sciences have studied the relationship between the changes in the bitcoin prices and the media attention to this cryptocurrency. The researchers examined the mentions of bitcoin in the media between 2017 and 2021 and built a mathematical model that revealed the strong relationship between media attention and bitcoin prices. The study was published in the Applied Stochastic Models in Business and Industry journal.

The question of bitcoin price modeling fascinates many researchers. Models for describing the dynamics of cryptocurrencies are used in many areas — for example, when predicting bubbles or diversifying investment portfolios. At the same time, the dynamics of the bitcoin price is fundamentally different from the dynamics of many other financial instruments, since both the bitcoin price and its volatility change very sharply (the so-called jump-type dynamics). This is the reason why researchers are constantly looking for new and more sophisticated approaches to modeling. Forecasting the bitcoin price based on the corresponding media attention may be one of these approaches.

Two researchers from the HSE Laboratory of Stochastic Analysis and its Applications, Vladimir Panov and Ekaterina Morozova, have proposed a new model for studying the relationship between media attention to bitcoin and changes in its price. This model is based on the class of Lévy processes, which is widely used to describe a time series with discontinuous dynamics. The paper considers data on the mention of bitcoin in the media for five years — from 2017 to 2021. The popularity of the topic was measured by Google Trends indicators.

 

The big difficulty is that media attention and bitcoin returns have a mutual effect on each other — a sharp change in returns can significantly increase media attention and, conversely, statements by well-known politicians and other news can lead to a sharp rise or fall in the price of bitcoin. This problem was solved with the help of new methods of low-frequency data analysis, which were developed specifically for models based on the Lévy processes.

Ekaterina Morozova
Research Assistant, Laboratory of Stochastic Analysis and its Applications

An interesting question is whether we can induce an increase in the cost of bitcoin by using publications in the media, that is, by increasing PR activity? Most likely, we can. At the same time, it’s important to understand that the exact impact of the PR strategy is actually unknown to us, since all models are stochastic, that is, based on probabilistic approaches.

Vladimir Panov
Associate Professor, Faculty of Economic Sciences

The results of this study and the proposed model help in studying the dynamics of the cryptocurrency market and the influence of external factors on it. This is undoubtedly useful for further academic research, policy development and risk assessment. This work is also valuable for investors and can help them make more informed decisions regarding trading strategies, risk management and investment timing.
IQ

February 13