Using correlated stochastic differential equations to forecast cryptocurrency rates and social media activities

The growing interconnectivity of socio-economic systems requires one to treat multiple relevant social and economic variables simultaneously as parts of a strongly interacting complex system. Here, we analyze and exploit correlations between the price fluctuations of selected cryptocurrencies and so...

Full description

Saved in:
Bibliographic Details
Published in:Applied network science 2020-03, Vol.5 (1), p.1-30, Article 17
Main Authors: Dipple, Stephen, Choudhary, Abhishek, Flamino, James, Szymanski, Boleslaw K., Korniss, G.
Format: Article
Language:English
Subjects:
Blockchain and cryptocurrencies
Brownian motion
Complex systems
Complexity
Computer Appl. in Social and Behavioral Sciences
Computer Science
Correlation analysis
Cryptocurrency
Differential equations
Differential geometry
Digital currencies
Digital media
Mathematical models
Maximum likelihood estimation
Noise prediction
Ornstein-Uhlenbeck process
Parameter estimation
Predictive modeling
Simulation and Modeling
Social factors
Social media activity
Social networks
Stochastic differential equations
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The growing interconnectivity of socio-economic systems requires one to treat multiple relevant social and economic variables simultaneously as parts of a strongly interacting complex system. Here, we analyze and exploit correlations between the price fluctuations of selected cryptocurrencies and social media activities, and develop a predictive framework using noise-correlated stochastic differential equations. We employ the standard Geometric Brownian Motion to model cryptocurrency rates, while for social media activities and trading volume of cryptocurrencies we use the Geometric Ornstein-Uhlenbeck process. In our model, correlations between the different stochastic variables are introduced through the noise in the respective stochastic differential equation. Using a Maximum Likelihood Estimation on historical data of the corresponding cryptocurrencies and social media activities we estimate parameters, and using the observed correlations, forecast selected time series. We successfully analyze and predict cryptocurrency related social media and the cryptocurrency market itself with a reasonable degree of accuracy. In particular, we show that our method has impressive accuracy in predicting whether a cryptocurrency market will increase or decrease a day in the future, a significant result with regards to investing and trading cryptocurrencies.
ISSN:2364-8228
2364-8228
DOI:10.1007/s41109-020-00259-1