Forecasting cryptocurrencies under model and parameter instability

This paper studies the predictability of cryptocurrency time series. We compare several alternative univariate and multivariate models for point and density forecasting of four of the most capitalized series: Bitcoin, Litecoin, Ripple and Ethereum. We apply a set of crypto-predictors and rely on dynamic model averaging to combine a large set of univariate dynamic linear models and several multivariate vector autoregressive models with different forms of time variation. We find statistically significant improvements in point forecasting when using combinations of univariate models, and in density forecasting when relying on the selection of multivariate models. Both schemes deliver sizable directional predictability.

[1]  J. Stock,et al.  A Comparison of Direct and Iterated Multistep Ar Methods for Forecasting Macroeconomic Time Series , 2005 .

[2]  F. Diebold,et al.  Comparing Predictive Accuracy , 1994, Business Cycles.

[3]  David Madigan,et al.  Dynamic Logistic Regression and Dynamic Model Averaging for Binary Classification , 2012, Biometrics.

[4]  Nicholas G. Polson,et al.  Sequential Learning, Predictability, and Optimal Portfolio Returns: Sequential Learning, Predictability, and Optimal Portfolio Returns , 2014 .

[5]  Dimitris Korobilis,et al.  Large Time-Varying Parameter VARs , 2012 .

[6]  S. Nadarajah,et al.  Statistical Analysis of the Exchange Rate of Bitcoin , 2015, PloS one.

[7]  N. Barberis.  Investing for the Long Run When Returns are Predictable , 2000 .

[8]  Francis X. Diebold,et al.  The Rodney L. White Center for Financial Research Financial Asset Returns, Direction-of-Change Forecasting and Volatility , 2003 .

[9]  Stig Vinther Møller,et al.  Forecasting House Prices in the 50 States Using Dynamic Model Averaging and Dynamic Model Selection , 2014 .

[10]  Leopoldo Catania,et al.  Dynamic Model Averaging for Practitioners in Economics and Finance: The eDMA Package , 2016, ArXiv.

[11]  M. Halling,et al.  Predictive Regressions with Time-Varying Coefficients , 2008 .

[12]  Nicholas G. Polson,et al.  Sequential Learning, Predictability, and Optimal Portfolio Returns , 2013 .

[13]  Leopoldo Catania,et al.  Portfolio optimisation under flexible dynamic dependence modelling , 2016, Journal of Empirical Finance.

[14]  G. Koop,et al.  Forecasting In ation Using Dynamic Model Averaging , 2009 .

[15]  G. Koop,et al.  UK Macroeconomic Forecasting with Many Predictors: Which Models Forecast Best and When Do They Do So? , 2009 .

[16]  Adrian E. Raftery,et al.  Prediction under Model Uncertainty Via Dynamic Model Averaging : Application to a Cold Rolling Mill 1 , 2008 .

[17]  Gianni Amisano,et al.  Comparing Density Forecasts via Weighted Likelihood Ratio Tests , 2007 .

[18]  Ľuboš Pástor.  Portfolio Selection and Asset Pricing Models , 1999 .

[19]  Florian Huber,et al.  Predicting crypto-currencies using sparse non-Gaussian state space models , 2018, 1801.06373.

[20]  J. Geweke,et al.  Comparing and Evaluating Bayesian Predictive Distributions of Asset Returns , 2008 .

[21]  R. Stambaugh,et al.  Predictive Regressions , 1999 .

[22]  Peter Reinhard Hansen,et al.  The Model Confidence Set , 2010 .

[23]  Michael A. West,et al.  Time Series: Modeling, Computation, and Inference , 2010 .

[24]  Helmut Ltkepohl,et al.  New Introduction to Multiple Time Series Analysis , 2007 .

[25]  Angelika I. Kokkinaki,et al.  Bitcoin Is Volatile! Isn't that Right? , 2014, BIS.

[26]  Siem Jan Koopman,et al.  Time Series Analysis by State Space Methods , 2001 .