HODL Waves and Coin Age: Reading the Age Distribution of Crypto Supply

Key Takeaways

• What it is: The current supply split into bands by time since each unit last moved, expressed as shares of one total.
• The right technique: Forecast the band populations under zero activity, then compare the observed populations against that forecast. The difference is the behaviour; the forecast is the drift.
• The realized-cap weighted version is a different chart: Weighting bands by cost basis rather than by units answers a different question and can point the opposite way.
• Age is model-dependent: It is a native property of an output on a UTXO chain and a reconstructed estimate on an account chain, where the provider must impose an accounting convention first.

Who This Guide Is For

This guide is for anyone who has looked at a HODL waves chart, seen the long-held bands expanding, and concluded that holders were accumulating. That conclusion may be right, but the chart alone cannot support it, because the same expansion happens automatically when nothing occurs. It is also for anyone comparing HODL waves across two providers or two networks, where band edges and age accounting differ enough to make the charts incomparable without adjustment.

Educational content about how a chart is built. Not individualized financial advice. Every figure is hypothetical.

What Does HODL Waves Measure?

HODL waves measures the composition of current supply by coin age, a unit's age being the time since it last moved on the ledger. Every band is a share of one denominator, which is why the visual is a stacked area filling the chart's full height.

Coin age is not holding period

Age counts time since the last on-chain movement, not time under one owner. A holder who has never sold but consolidated wallets last week appears in the youngest band. Coins bought and sold repeatedly inside an exchange never move on-chain and keep ageing where they sat at deposit. The metric measures ledger stillness, which overlaps with conviction only loosely.

The composition constraint

Because the bands partition one total, they carry less independent information than their number suggests. If the 1-to-3-month band gains three percentage points, the other bands have lost exactly three between them. Commentary citing a swelling old band and a shrinking young band as two confirming observations is citing one observation with two labels.

The denominator moves too

On a chain with ongoing issuance, total supply grows, so every band's share falls unless its units grow at least as fast, and newly issued coins land in the youngest band by definition. Both effects push the young band up and every other share down independently of anything holders do, and neither is visible in a chart of shares.

How Is HODL Waves Constructed?

The build is short, and each step introduces a convention that has to be published for the chart to be reproducible.

1. Enumerate supply as lots. On a UTXO chain they already exist as unspent outputs. On an account chain they must be synthesized.
2. Age each lot. Age is the difference between the current block height and the creating block height, converted to elapsed time. The Bitcoin Developer Guide: Block Chain describes the height and confirmation structure this rests on.
3. Assign to a band. Band edges are a provider convention, not a protocol rule.
4. Divide by total supply. Which supply counts (all issued, circulating, entity-adjusted) decides the denominator for every band at once.
5. Optionally reweight by realized capitalization instead of by units, which produces a related but distinct chart.

The UTXO dependence of coin age

On a UTXO chain, age is native. Each unspent output records the block that created it, so a lot's age is read directly, and the Bitcoin Developer Guide: Transactions sets out the structure making that possible. An account chain differs in kind, not degree. An account holds one integer balance with no record of which units arrived when, so sending 3 units out of 10 does not identify which 3 left. The provider must impose first-in-first-out, last-in-first-out, or proportional ageing before any band populations exist, and the three produce genuinely different HODL waves from identical chain data. See the UTXO versus account model guide.

Pooled contract balances

A liquidity pool, lending market, staking contract, or bridge escrow holds one balance for many depositors. Every deposit and withdrawal touches it, so the pooled coins stay perpetually young no matter how long individual depositors have been committed. Where a large share of supply sits in such contracts, the young bands are inflated by protocol mechanics rather than holder turnover.

Where band edges come from

Edges are chosen for legibility. A split at one month, three months, six months, one year, two years, and beyond is a design decision, and a provider using different cut points produces a chart that cannot be compared band-for-band. Edge choice also sets how much mechanical drift is visible, since narrow bands empty and refill faster than wide ones.

Formula and Measurement Logic

Band share = (units whose age falls inside the band) / (total supply counted), with the bands partitioning all supply so the shares sum to 1

The interesting arithmetic is the drift, which is fully predictable when nothing happens. Over d days with no spending, every unit's age increases by exactly d. For a band of width w days holding a roughly uniform spread of ages, the fraction crossing into the next band up is d / w, and the band receives an inflow from below on the same rule.

Three consequences follow directly.

• Narrow bands turn over fast. A band 30 days wide empties completely over 30 quiet days; a band 360 days wide loses one twelfth. Charts with fine young bands and coarse old bands therefore show most of their motion at the young end regardless of behaviour.
• The oldest band is absorbing. It has no upper edge, so it only ever receives. Under zero activity it grows every single day, and the coins in it that are permanently lost can never leave.
• The youngest band is the only one that needs transactions. Nothing ages into it. Only spending or new issuance refills it, which is why a collapsing young band is the least mechanical signal on the chart.

How Should HODL Waves Be Interpreted?

Interpret HODL waves by first computing what the bands would look like if nothing happened, then reading only the gap between prediction and reality. Everything else on the chart is the calendar.

Building the zero-activity baseline

Take the band populations at the start of the period, age them forward by the elapsed days using the width rule above, and compare predicted against observed. A band that grew more than predicted received coins from spending; one that grew less lost coins to spending. This residual is the behavioural content, and it is usually far smaller than the raw band movement.

What the residual can and cannot say

• It identifies which age cohorts spent. Coins leaving a band early appear in the youngest band, so a shortfall in an old band paired with an excess in the youngest locates the spending by age.
• It does not identify who. One custodian reorganizing cold storage produces the same residual as thousands of independent holders acting alike.
• It does not identify why. Age says nothing about price, so a reset from an old band to the youngest is equally consistent with a sale, a custody migration, or a wallet upgrade.

The two-band special case

Splitting supply at a single age edge gives a long-term and a short-term holder supply series: HODL waves with two bands and the same mechanical drift. See the long-term holder supply guide and the short-term holder supply guide. Everything here about ageing across an edge applies to them unchanged.

Step-by-Step Workflow

1. Record the provider's band edges and confirm they were not revised inside the period.
2. Confirm whether age is native output age or reconstructed under an account-model convention.
3. Confirm the supply denominator and whether issuance is material over the period.
4. Age the starting band populations forward by the elapsed days to build a zero-activity baseline.
5. Subtract the baseline from the observed end state and work only with the residual.
6. Check the youngest band separately, since it is the only band that cannot be filled by ageing.
7. Check whether one large entity's movement explains the residual before calling it holder behaviour.
8. Check whether the realized-cap weighted version agrees with the unit-weighted one.

Worked Hypothetical Scenario

All figures here are invented and describe no real network. Assume ages are spread evenly inside each band and the chain has a fixed supply of 1,000,000 units with no issuance. The left half of the table is the starting distribution across six age bands; the right half is the same supply after 30 days in which not one transaction occurs.

Where each number comes from

• The 0-to-30 band is 30 days wide, so all 120,000 units age past its upper edge and land in the 30-to-90 band. Nothing refills it, because refilling requires a transaction.
• The 30-to-90 band is 60 days wide, so 30/60, or half of its 180,000 units, crosses into 90-to-180. It keeps 90,000 and receives 120,000 from below, ending at 210,000.
• The 90-to-180 band is 90 days wide, so 30/90, or one third of 150,000, moves up. It keeps 100,000 and receives 90,000, ending at 190,000.
• The 180-to-360 band is 180 days wide, so one sixth of 180,000, or 30,000, moves up. It keeps 150,000 and receives 50,000, ending at 200,000.
• The 360-to-720 band is 360 days wide, so one twelfth of 240,000, or 20,000, moves up. It keeps 220,000 and receives 30,000, ending at 250,000.
• The oldest band has no upper edge, so it keeps all 130,000 and receives 20,000, ending at 150,000.

What a reader would say about this chart

Supply held for more than a year went from 370,000 units to 400,000, or from 37.0 percent to 40.0 percent, a rise of three percentage points in a month. The under-one-month band went from 12.0 percent to zero. Every band above one month grew. Read as a chart, this is a textbook picture of long-term accumulation and vanishing short-term supply.

No coin moved. No holder made a decision. No capital entered or left. The entire chart is 30 days of calendar arithmetic applied to a static distribution, and it looks like conviction only because ageing and accumulation produce the same shape. Without a zero-activity baseline, a quiet month and an accumulation month are indistinguishable.

What Can Make the Interpretation Wrong?

• Reading mechanical ageing as accumulation. The central error, producing exactly the shape people look for.
• Treating bands as independent evidence. They are shares of one total, so a rising old band and a falling young band are one observation.
• Ignoring the youngest band's asymmetry. It is the only band that cannot be refilled by ageing, so it collapses during quiet periods by construction.
• Forgetting issuance. New supply enters the youngest band and enlarges the denominator, pushing every other band's share down without any holder acting.
• Comparing charts with different band edges. A three-month band and a two-month band hold different populations and cannot be read against each other.
• Overlooking custodial resets. One exchange reorganizing cold storage moves a large block from an old band to the youngest in a day. That looks like mass capitulation and is a wallet migration.
• Assuming age tracks a person. Coins traded repeatedly inside an exchange never move on-chain and keep ageing, while a holder who never sold appears young after consolidating wallets.

Cross-Network and Provider Comparison

On one family of chains HODL waves is measured. On the other it is estimated.

UTXO chains give coin age away. Every unspent output names its creating block, so ageing supply is a lookup and two competent providers should agree closely, differing only on band edges and the supply denominator. Account chains have no lot structure, so the provider chooses an accounting convention and the chart inherits it. First-in-first-out ageing makes a spending account look older than last-in-first-out does from the same history, and neither is more correct, because the underlying fact does not exist on the ledger.

Contract-heavy networks add a structural bias toward youth. Coins inside a pooled contract are touched by every interaction with it, so a chain where much supply sits in automated market makers, lending markets, or staking contracts shows a young age profile reflecting protocol design rather than holder behaviour. Whether a staking deposit freezes or resets a coin's age is a provider convention worth checking on proof-of-stake networks, where the affected share can be large.

Between providers on the same chain, check band edges first, the supply denominator second, entity adjustment third. Those three account for most apparent disagreements before data quality arises.

Advanced Analytical Methods

Realized-cap weighted HODL waves

Weighting each band by its contribution to realized capitalization rather than by unit count answers a different question: not what share of the coins is this old, but what share of the network's aggregate cost basis was laid down in this age band. Old coins usually carry a very low basis, so they can dominate the unit-weighted chart while barely registering in the value-weighted one. Continuing the hypothetical above with an average last-move price per band, the two weightings diverge sharply.

Realized-cap shares are rounded and sum to 99.9 percent. The oldest band holds 15 percent of the coins and 0.8 percent of the aggregate cost basis; the two youngest hold 40 percent of the coins and 68.2 percent of the basis. A chart of units says most supply is old and settled. A chart of value says most of the money at stake was committed within the last six months. Both are correct, so neither confirms the other.

Publish the residual, not the bands

Charting the observed band populations minus the zero-activity forecast turns HODL waves from a chart dominated by drift into a chart of behaviour. It is straightforward if the underlying age distribution is available, and it removes the largest misreading of the metric.

Band-edge sensitivity

Recompute the chart under a second set of edges. Conclusions surviving both are about the supply; conclusions appearing under one and vanishing under the other are about the edges.

Entity-adjusted ageing

Suppressing age resets caused by identified internal transfers stops a custodial reorganization from dumping a large block into the youngest band. It is the most valuable single filter on the chart, and it materially changes the young bands where custody is concentrated.

Practical Checklist

• I recorded the band edges and confirmed they did not change in the period.
• I know whether age is native output age or reconstructed under an account convention.
• I built a zero-activity baseline before reading any band movement.
• I treated the youngest band separately, since ageing cannot refill it.
• I did not cite a rising old band and a falling young band as two pieces of evidence.

Frequently Asked Questions

References

These sources should be reviewed during editorial verification. They support data structures and methods, not the hypothetical conclusion. Provider formulas, chain rules, and APIs can change. Confirm current documentation before publication.

• Bitcoin Developer Guide: Transactions: the output structure that makes coin age a native, readable property.
• Bitcoin Developer Guide: Block Chain: block height and confirmations, from which elapsed age is derived.
• Ethereum.org: Technical Introduction: the account and state model that forces an ageing convention.
• Glassnode: Metric Catalog: where HODL wave bands and age-cohort supply series are catalogued.
• Glassnode: Realized Capitalization Guide: the weighting used by the realized-cap version of the chart.
• Coin Metrics: Network Data Glossary: supply and ledger definitions that set the denominator.