Risk Aversion and Cryptocurrency Price Prediction

Abstract

This study proposes a method for constructing cryptocurrency portfolios using forecast models. We predict returns on four liquid cryptocurrencies (Bitcoin, Litecoin, Ripple, and Dash) and determine the weights on the cryptocurrencies based upon a dynamic allocation framework. We assess the performances of the portfolios using the performance fee measure. Our results present that the proposed portfolios outperform the benchmark portfolio with the conventional level of the risk aversion parameter. The economic gain for an investor is equivalent to 12% per week. The economic gain is sensitive to a change in the risk aversion parameter, which contrasts with the studies of exchange rates. This is due to high volatility on the cryptocurrencies. Our predictors are related to the price momentum effects and they outperform widely used network factors.

Keywords: Cryptocurrency, Bitcoin, Portfolio evaluation, Forecast model, Risk aversion

JEL Classification: G10, G11, G17