Article,
Analyzing swings in Bitcoin returns: a comparative study of the LPPL and sentiment-informed random forest models
Affiliations
- [1] International Institute for Management Development [NORA names: Switzerland; Europe, Non-EU; OECD];
- [2] Moonpass, Berlin, Germany [NORA names: Germany; Europe, EU; OECD];
- [3] WHU - Otto Beisheim School of Management [NORA names: Germany; Europe, EU; OECD];
- [4] Copenhagen Business School [NORA names: CBS Copenhagen Business School; University; Denmark; Europe, EU; Nordic; OECD]
Abstract
Forecasting Bitcoin’s returns continues to be a challenging endeavor for both scholars and practitioners. In this paper, we train a random forest model on a variety of features, with the aim of predicting pronounced changes in the returns of Bitcoin. The model that we present in this paper outperforms the baseline model with which we compare it: the LPPL model. Our results have implications for scholars studying financial prediction models, as well as for practitioners interested in Bitcoin investment.
Keywords
Bitcoin,
Bitcoin investment,
Bitcoin returns,
LPPL,
LPPL model,
baseline,
baseline model,
changes,
comparative study,
features,
financial prediction models,
forecasting,
forest model,
investment,
model,
practitioners,
prediction model,
pronounced changes,
random forest model,
results,
return,
scholars,
study,
swing