Assessing Bitcoin Return Extrema in the Context of Extreme Value Theory

Abstract

This study examines the distribution of extreme Bitcoin returns within the framework of Extreme Value Theory. Extreme Value Theory is crucial for modeling rare but impactful events in fields such as climate science, finance, insurance, engineering, and environmental sciences, among others. Using traded price data from Binance, the world’s largest cryptocurrency exchange by volume, we analyze weekly extreme values (both maxima and negative minima) of hourly Bitcoin returns from January 2018 to July 2024. Specifically, we explore both tails of the Bitcoin hourly return distribution using the generalized extreme value class of distributions. Our results suggest that both tails can be effectively modeled with the generalized extreme value framework. This enables us to make hourly return predictions over weekly forecast horizons which can be instrumental in conducting risk assessments. Additionally, these insights could enhance the development of trading algorithms for this increasingly popular asset class, which is integrated already into traditional capital markets via exchange-traded funds (ETFs) on major stock exchanges. © 2025 Elsevier B.V., All rights reserved.