Forecasting Cryptocurrency Prices using Contextual ES-adRNN with Exogenous Variables

TL;DR

The paper addresses the challenge of forecasting cryptocurrency prices amid high volatility and external influences by introducing a hybrid contextual model, cES-adRNN, that combines exponential smoothing with a dilated, attention-based RNN and a separate context track. This architecture leverages exogenous variables and a dynamic context derived from representative series to improve multivariate forecasting through cross-learning and per-series modulation. Extensive experiments on 15 cryptocurrencies with 17 input series demonstrate that the exogenous-driven cES-adRNN+ achieves the lowest forecasting errors across horizons of $1$, $7$, and $28$ days, outperforming both statistical and ML baselines; Giacomini-White tests corroborate the model's superior conditional predictive ability. The work highlights the practical value of integrating exogenous data and contextual information for reliable crypto price forecasts and suggests favorable deployment prospects with offline training and periodic retraining.

Abstract

In this paper, we introduce a new approach to multivariate forecasting cryptocurrency prices using a hybrid contextual model combining exponential smoothing (ES) and recurrent neural network (RNN). The model consists of two tracks: the context track and the main track. The context track provides additional information to the main track, extracted from representative series. This information as well as information extracted from exogenous variables is dynamically adjusted to the individual series forecasted by the main track. The RNN stacked architecture with hierarchical dilations, incorporating recently developed attentive dilated recurrent cells, allows the model to capture short and long-term dependencies across time series and dynamically weight input information. The model generates both point daily forecasts and predictive intervals for one-day, one-week and four-week horizons. We apply our model to forecast prices of 15 cryptocurrencies based on 17 input variables and compare its performance with that of comparative models, including both statistical and ML ones.

Figures (5)

• Figure 1: Prices of cryptocurrencies.
• Figure 2: Block diagram of the forecasting model.
• Figure 3: Sliding windows for generating training data.
• Figure 4: Architecture of adRNN (elements in red do not exist in the context track).
• Figure 5: Examples of forecasts for BTC: top panel - $h=1$, middle panel - $h=7$, bottom panel - $h=28$.