Authors:
Leandro Maciel
1
;
Rosangela Ballini
2
and
Fernando Gomide
3
Affiliations:
1
Department of Business Administration, Faculty of Economics, Business and Accounting, University of São Paulo, São Paulo, Brazil
;
2
Department of Economic Theory, Institute of Economics, University of Campinas, São Paulo, Brazil
;
3
Department of Computer Engineering and Automation, School of Electrical and Computer Engineering, University of Campinas, São Paulo, Brazil
Keyword(s):
Data Driven Fuzzy Modeling, Cryptocurrency, Forecasting.
Abstract:
The paper develops a data-driven fuzzy modeling procedure based on level set to forecast cryptocurrencies prices. Data-driven level set is a novel fuzzy modeling method that differs from linguistic and functional fuzzy models in how the fuzzy rules are built and processed. The level set-based model outputs the weighted average of output functions associated with the fuzzy rules. Output functions map the activation levels of the fuzzy rules directly in the model outputs. Computational experiments are done to evaluate the level set method to forecast the closing prices of Bitcoin, Ethereum, Litecoin and Ripple. Comparisons are made with ARIMA, ETS, MLP and naı̈ve random walk. The results suggest that the random walk outperforms most methods addressed in this paper, but it is surpassed by the level set model for Ethereum. When performance is measured by the direction of price change, the level set-based fuzzy modeling performs best amongst the remaining methods.