Onchain analysts group Bitcoin addresses into size bands and give them marine nicknames. Shrimp hold less than one coin, crabs one to ten, fish ten to a hundred, sharks a hundred to a thousand, whales a thousand to ten thousand, humpbacks above that. The chart of who is accumulating and who is distributing is one of the most shared in the space. It is also one of the most misunderstood, for a reason that sits in the data itself.
What the cohorts measure
Each band tracks the total supply held by addresses in that size range, and how that total changes. A rising shrimp balance means small holders are accumulating in aggregate. A falling whale balance means large holders are net distributing. Both Glassnode and CryptoQuant publish versions, though the band definitions vary slightly.
The appeal is obvious. It appears to answer the question everyone wants answered, which is whether smart money is buying. The problem is that the question assumes wallet size maps onto sophistication, and it does not.
The labelling problem
An address is not a person. Three distortions run through every cohort chart.
Exchanges and custodians hold enormous balances in a small number of addresses. When Coinbase moves inventory between wallets, the whale and humpback bands move, and no investor did anything. Serious providers exclude known exchange addresses, but the exclusion is only as complete as the labelling.
ETFs made this worse. Spot ETF holdings sit in custodial addresses that look like whales. Thousands of retail investors buying shares of a fund shows up in the data as one very large entity accumulating. The economic reality is the opposite of what the chart appears to say.
Individuals also split holdings across many addresses for privacy and security, which quietly shifts one holder down into smaller bands. The reverse happens when a wallet consolidates.
Entity-adjusted data
The partial fix is heuristic clustering, grouping addresses that appear to be controlled by the same entity based on how they spend together. Entity-adjusted cohort data is meaningfully better than raw address data and is what serious desks use. It is still an estimate, it still fails on well-managed privacy practices, and it still cannot see inside a custodian. Better, not solved.
What the cohorts are actually good for
Used carefully, two things. First, divergence. When small cohorts accumulate while large ones distribute over a sustained period, that is a real behavioural split worth noting, even if the identities behind it are fuzzy. Second, extremes. Cohort behaviour at cycle turning points has historically been distinctive, with small holders arriving late and large holders distributing into strength.
What they are not good for is a weekly signal. The bands move slowly, the noise from custodial reshuffling is large relative to real changes, and the temptation to narrate every wiggle is strong. For the holding question, long-term holder supply and HODL waves slice by coin age rather than wallet size, which sidesteps the labelling problem almost entirely and is usually the better tool.
How the bands are drawn
The size bands themselves are a convention, not a natural feature of the data. Different providers draw the boundaries in slightly different places, which means a chart of whale accumulation from one source and another from a second source can disagree simply because a cluster of addresses sits either side of a line. There is nothing special about one thousand coins other than that it is a round number.
The bands also do not adjust for price. An address holding one coin was a small holder in 2015 and is a small holder now, but the dollar value of that position has changed by orders of magnitude, and so has the type of person likely to hold it. Long-run comparisons across the bands quietly compare different populations, which is one more reason the age-based metrics tend to be the sounder tool.
Common questions
How many whales are there? Nobody knows. Counts of whale addresses are counts of addresses, and one entity can hold many or one address can hold many entities’ coins.
Are whales smarter than retail? On average larger holders have held longer, which correlates with better outcomes. That is survivorship as much as skill, and it says nothing about any individual trade.
Which is better, cohorts or coin age? Coin age, for most questions. It measures behaviour directly rather than inferring it from balance size. See the metrics guide for the full set.
Research and education only, not financial advice.
Leave a Reply