A portfolio built around a cryptocurrency adoption model had a 96.85% monthly standard deviation in the researchers’ historical test, alongside an 18.20% mean monthly excess return. The researchers found that the model’s innovator component predicted later returns, while total active-address growth did not show a statistically significant signal.
Khaladdin Rzayev, Athanasios Sakkas and Andrew Urquhart authored the paper, published May 16, 2025. Its sample spans January 2014 to July 2022 and contains monthly observations of active addresses across 77 cryptocurrencies. The model split those address trends into “innovator” and “imitator” components, both statistical estimates rather than labels attached to particular wallets or investor groups.
In the return analysis, the innovator estimate was the only component with statistically significant predictive power. The imitator estimate and overall active-address growth did not reach that threshold.
That distinction sets a limit on what the study establishes. Total active-address growth is a broad activity measure; “innovator adoption” is a narrower model output. The findings do not show that a rise in address counts for a specific cryptocurrency will come before price appreciation. Nor do they identify real first movers or test whether those investors gained an edge over other market participants.
The researchers also built a long-short portfolio using the model’s score, going long on the top 30% of assets and short on the bottom 30%. Monthly standard deviation was 96.85%, against an 18.20% mean monthly excess return. This spread indicates considerable variation in the test’s monthly results. Standard deviation measures how widely outcomes vary around their average; the reported figure was high relative to the excess return.
These statistics describe a retrospective exercise, not a projection for an individual investor. The results do not establish that a trader could have identified the same asset groups as the model in real time. The analysis also did not compare its adoption signals with market-attention measures.
The observations end in July 2022, so the study does not establish whether the relationship persisted under later market conditions.
