How Long-Term Holders Actually Behave On-Chain
Coin age data offers one of the few objective windows into holder behaviour. It is also routinely oversold. Here is what it can and cannot show.
Most claims about investor psychology are unfalsifiable. On-chain data is an exception, because coins carry a timestamp of when they last moved.
That makes it genuinely useful and also heavily misused.
What the data actually measures
Every unspent output records when it was created. Aggregate that and you can see the distribution of how long coins have been sitting still.
Analysts typically split this into coins that have not moved for some threshold, often around five months, and those that have. The first group is labelled long-term holders, the second short-term.
The behaviour of those two groups across cycles is consistent enough to be worth knowing.
The pattern that repeats
During declines, the long-held supply grows. Coins move from recent buyers to holders who do not sell. This has happened in every bear market on record.
During advances, the long-held supply shrinks. Old coins move, meaning people who held through the decline sell into strength.
The handover is gradual, not sudden. Distribution from long-term holders begins well before a peak and continues through it, which is why “smart money is selling” headlines tend to be several months early.
This is not a prediction system. It is a description of what has already happened, which is the honest thing to say about all on-chain analysis.
What the data cannot tell you
Who owns anything. A large address may be a custodian holding coins for a hundred thousand customers. Treating it as one investor is a category error, and a substantial share of “whale” analysis makes it.
Intent. A coin moving is not a coin selling. It may be a custody migration, an internal transfer, a rebalance between an investor’s own wallets, or collateral being posted.
Anything about the price tomorrow. The correlations are suggestive over months and useless over days.
The activity of institutional holders. Coins inside an ETF’s custody structure barely move. Enormous changes in beneficial ownership produce almost no on-chain footprint.
That last point is the one that undermines a lot of older analysis. As more supply sits in structures where ownership changes without transactions, the chain sees less and less of what actually happens.
The metrics worth understanding
| Metric | What it shows | Main limitation |
|---|---|---|
| Coin age distribution | How long supply has been still | Custodial addresses distort it |
| Supply last active over a year | The slowest-moving cohort | Includes lost coins permanently |
| Realised price | Average cost basis of moved coins | Says nothing about unmoved supply |
| Exchange balances | Coins on venue addresses | Address labelling is imperfect |
The lost-coin problem deserves emphasis. A meaningful share of the supply that has not moved in a decade has not moved because the keys are gone. Every metric that counts dormant supply as patient holders is counting graves as residents.
How I use it
As context, at a timescale of quarters, and never as a signal.
Knowing that long-held supply is growing during a decline tells me the decline is being absorbed rather than compounded. That is genuinely reassuring and not remotely actionable.
Anyone presenting a coin-age chart with an arrow and a price target is selling something. The data supports the first half of that sentence and not the second.
Where to look
The metrics are published by several analytics firms, and the raw data is on the chain for anyone willing to compute it. Exchange balance figures should be treated with the most caution, since address labelling relies on heuristics and large venues such as the venue I buy through periodically reorganise their wallet structure, which produces apparent flows that are nothing of the kind.
This is a personal account of holding through market cycles. It describes what one person did and why. It is not a recommendation, and past cycles do not predict future ones.