Datsenko N. Model system for evaluating and forecasting innovative financial assets (cryptocurrencies)

The dissertation is devoted to the further development of theoretical basis and experimental tools for modeling and forecasting dynamics of innovative financial assets (cryptocurrencies). The thesis is defined the role and place of cryptocurrencies in the current level of the world economy progress, clarifies the categorical apparatus and economic essence of the definition of «cryptocurrency», developed taxonomy of cryptocurrencies on the basis of their technological and cryptographic properties. A conceptual approach to the evaluation and short-term forecasting of crypto-assets based on the CRISP-DM data mining inter-industry standard is proposed. The binary autoregression tree algorithm (BART) is proposed. This model performs piecewise linear approximation phase space and build an autoregressive model for each segment. Numerical estimation of the forecast horizon was calculated using fractal and entropy analysis, and cryptocurrencies time series forecasts were performed using BART and ARIMA-ARFIMA models. The conducted theoretical and experimental study showed the adequacy of the built models and the effectiveness of their application in order to support making decisionson on the crypto market.