When the numbers don't add up: anomalies and inconsistencies in bitcoin prices and daily data

Despite Bitcoin's immutable public ledger, many studies have poor replicability due to inconsistencies in the way data are collected and stored; furthermore, several important metrics relating to the mempool and Lightning Network must be captured in real-time, and many valuable data series have missing observations. An attempt to replicate and extend two well-cited papers is documented. This paper may also serve as a reference for relative newcomers to Bitcoin from the fields of statistics, business, and econometrics, due to its low level approach to data. Information and literature related to protocol changes, price discovery, and the computation of daily averages is summarized. Research gaps are identified; best data practices are suggested. Block-level data, computed in native units, provides a superior perspective of Bitcoin activity to daily fiat metrics. Readers benefit from an improved understanding of (i) valid approaches and data sources, (ii) some of Bitcoin's quirks (there are many), and (iii) potential research pitfalls.