Predicting crypto‐currencies using sparse non‐Gaussian state space models

• Author: Florian Huber,Thomas Otto Zörner,Christian Hotz‐Behofsits
• Date: 01 September 2018
• Published date: 01 September 2018
• DOI: http://doi.org/10.1002/for.2524

RESEARCH ARTICLE

Predicting crypto‐currencies using sparse non‐Gaussian

state space models

Christian Hotz‐Behofsits

1

| Florian Huber

2

| Thomas Otto Zörner

2

1

Department of Marketing, Vienna

University of Economics and Business,

Vienna, Austria

2

Department of Economics, Vienna

University of Economics and Business,

Vienna, Austria

Correspondence

Florian Huber, Department of Economics,

Vienna University of Economics and

Business, Welthandelsplatz 1, 1020

Vienna, Austria.

Email: fhuber@wu.ac.at

Funding information

Czech Science Foundation, Grant/Award

Number: 17-14263S

Abstract

In this paper we forecast daily returns of crypto‐currencies using a wide variety of

different econometric models. To capture salient features commonly observed in

financial time series like rapid changes in the conditional variance, non‐normality

of the measurement errors and sharply increasing trends, we develop a time‐vary-

ing parameter VAR with t‐distributed measurement errors and stochastic volatil-

ity. To control for overparametrization, we rely on the Bayesian literature on

shrinkage priors, which enables us to shrink coefficients associated with irrelevant

predictors and/or perform model specification in a flexible manner. Using around

oneyearofdailydata,weperformareal‐time forecasting exercise and investigate

whether any of the proposed models is able to outperform the naive random walk

benchmark. To assess the economic relevance of the forecasting gains produced

by the proposed models we, moreover, run a simple trading exercise.

KEYWORDS

Bitcoin, density forecasting, stochastic volatility, t‐distributed errors

1|INTRODUCTION

In the present paper we develop a non‐Gaussian state‐

space model to predict the price of three crypto‐currencies.

Taking a Bayesian stance enables us to introduce shrinkage

into the modeling framework, effectively controlling for

model and specification uncertainty within the general

class of state‐space models. To control for potential outliers

we propose a time‐varying parameter vector autoregressive

(VAR) model (Cogley & Sargent, 2005; Primiceri, 2005)

with heavy‐tailed innovations,

1

as well as a stochastic

volatility specification of the error variances. Since the

literature on robust determinants of price movements in

crypto‐currencies is relatively sparse (for an example, see

Cheah & Fry, 2015), we apply Bayesian shrinkage priors

to decide whether using information from a set of potential

predictors improves predictive accuracy.

The recent price dynamics of various crypto‐currencies

point towards a set of empirical key features that an appro-

priate modeling strategy should accommodate. First, con-

ditional heteroskedasticity appears to be an important

regularity commonly observed (Chu, Chan, Nadarajah, &

Osterrieder, 2017). This implies that volatility is changing

over time in a persistent manner. If this feature is

neglected, predictive densities are either too wide (during

tranquil times) or too narrow (in the presence of tail

events, i.e., pronounced movements in the price of a

------------------------------------------------------- -- --- -- -- --- -- --- -- -- --- -- --- -- -- --- -- --- -- -- --- -- --- -- --- -- -- --- -- --- -- -

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the

original work is properly cited.

© 2018 The Authors Journal of Forecasting Published by John Wiley & Sons Ltd

1

For a recent exposition on how to introduce flexible error distributions

into VAR models with stochastic volatility, see Chiu, Mumtaz, and

Pintér (2017).

Received: 19 January 2018 Accepted: 11 February 2018

DOI: 10.1002/for.2524

Journal of Forecasting. 2018;37:627–640. wileyonlinelibrary.com/journal/for 627