Key Takeaways
- What it is: transfer volume over a period divided by a supply figure, borrowed loosely from the equation of exchange.
- The denominator is the whole argument: total supply, circulating supply, and free-float supply give three different velocities from one unchanged numerator.
- Much of supply provably cannot move: staking and vesting contracts, long-dormant cold storage, and coins whose keys are lost sit in the denominator and never appear in the numerator.
- Velocity can move with no behavioural change: if the staked share of supply rises while transfer volume is flat, free-float velocity rises as an arithmetic consequence.
- Not comparable across chains with different lock-up structures: a chain with half its supply staked and one with none can post identical circulating velocity while the coins actually able to move turn over at very different rates.
- Practical rule: state the supply definition beside every velocity figure, and check whether the numerator moved before attributing a change to behaviour.
Who This Guide Is For
Read this page if you have seen token velocity cited as evidence that a network is being used more or less, and want to know how much of that series is behaviour and how much is bookkeeping about the denominator.
Educational content. Not individualized financial advice.
What Does Token Velocity Measure?
Token velocity measures how many times, on average, a unit of supply would have to move to produce the observed transfer volume over a period. Nothing on chain reports a velocity; the number exists only once you choose what to divide by.
The idea is borrowed from the equation of exchange, usually written as money supply times velocity equals price level times real output. Applied to a blockchain, the money supply becomes coin supply and the transaction total becomes on-chain transfer volume. The borrowing is loose in three specific ways.
- The numerator is not final-goods spending. On-chain transfer volume includes exchange deposits, collateral movement, custody rebalancing, and self-payments. The macroeconomic original counts transactions in goods and services.
- There is no price level. The chain has no basket of goods priced in the asset, so the identity has no right-hand side to solve for.
- Supply is not a money stock. A money stock counts balances available for spending. A blockchain supply figure counts every unit in existence, including units that are provably immobile.
Plain-language definition
Velocity is the amount transferred in a period divided by how much of the asset exists. The second half of that sentence is where all the disagreement lives.
Technical definition
Velocity over an interval equals adjusted transfer volume during that interval divided by a chosen supply measure, usually annualised. The numerator must be an adjusted series, since raw volume includes self-payments and internal movement.
How Is Token Velocity Constructed?
The numerator decision is covered in the Swoopr Investment guide to on-chain transfer volume, so this section concentrates on the denominator, where velocity goes wrong most often.
Total supply
Total supply is every unit ever issued, less any provably destroyed, since issuance and burns are both recorded on chain. It is also the least meaningful denominator, since it includes vesting contracts that cannot be withdrawn yet, staking deposits that require an exit queue, and addresses that have not moved in a decade.
Circulating supply
Circulating supply attempts to remove units that are contractually unavailable: locked team and foundation allocations, unvested investor allocations, and in some definitions treasury holdings. That is an editorial judgment, and two providers can report circulating supply differing by tens of percent for a token with a complex vesting schedule.
Free-float supply
Free-float supply goes further and removes units that are observably immobile even though nothing contractually prevents their movement. Three categories dominate.
- Staked coins. On a proof-of-stake chain, coins in the deposit contract are locked by consensus rules and leaving requires an exit process, so they cannot appear in transfer volume while staked. Ethereum.org: Proof of Stake documents the deposit, activation, and exit mechanics.
- Long-dormant coins. Coins untouched for many years, identified by output age or last account activity. Some belong to patient holders and some belong to nobody, and the two are indistinguishable on chain.
- Provably lost coins. Units sent to addresses with no known spending key or to addresses that are demonstrably unspendable, permanently outside the numerator.
Free float best matches the economic question and is also the most estimated. Whether five years of inactivity means lost, held, or merely cold is unknowable from the ledger.
Formula and Measurement Logic
Velocity = adjusted transfer volume over the interval / supply
Free-float supply = circulating supply - staked and locked supply - long-dormant supply - provably lost supply
The second line is where disagreement comes from: every term on its right-hand side after the first is an estimate.
| Denominator | What it includes | How reliable it is | What the resulting velocity means |
|---|---|---|---|
| Total supply | Every unit issued, less burns | Close to a protocol fact | Turnover against the whole issued base, including units that cannot move |
| Circulating supply | Total, less contractually locked allocations | An editorial judgment; differs between publishers | Turnover against units nominally available |
| Free-float supply | Circulating, less staked, dormant, and lost | Heavily estimated; threshold-dependent | Turnover against units that could realistically move |
Velocity rises when the numerator rises or when the denominator falls, and those two causes are indistinguishable in the published series. Free float, being the smallest denominator, produces the largest velocity, so a chart with no stated definition cannot be compared even with itself across a methodology change.
Annualising by multiplying a daily velocity by 365 is a further trap. A trailing-year velocity computed from a trailing-year numerator avoids it.
How Should Token Velocity Be Interpreted?
Read a velocity change only after establishing which of its two terms moved. That single discipline removes most of the bad readings this metric produces. Four patterns cover almost every case.
- Numerator up, denominator flat. The genuine behavioural case, and the only pattern that supports a statement about usage.
- Numerator flat, denominator down. An arithmetic artifact. More coins entered staking, vesting locks, or long-term dormancy, so the same activity is measured against a smaller base.
- Numerator down, denominator down more. Velocity still rises while activity falls. Reported without decomposition, this reads as increasing usage during a decline.
- Both flat, definition changed. A provider switching from circulating to free-float supply produces a step change with no change in the world.
The popular claim that low velocity means holding and high velocity means spending survives only in the first case. In the others it describes the denominator. Worse, the metric can move opposite to the behaviour that caused it: a rise in staking lowers free float and raises free-float velocity, which looks like more activity while the underlying behaviour is people choosing to make their coins immobile.
Step-by-Step Workflow
- Write down the supply definition being used. If the source does not state one, treat the figure as uninterpretable rather than guessing.
- Confirm the numerator is an adjusted transfer series, not raw volume including change outputs and internal exchange movement.
- Confirm numerator and denominator are in the same units and cover the same interval.
- Plot the numerator and the denominator separately, above the ratio.
- Compute the same velocity against all three supply definitions and note the spread.
- Check the staked share of supply over the window, since a change there moves free-float velocity on its own.
- Check for vesting cliffs or unlock events, which move circulating supply in steps.
- Before any cross-chain comparison, align the supply definition and check the lock-up structures.
Worked Hypothetical Scenario
Every figure below is hypothetical and chosen so the divisions are checkable by hand.
Part one: one numerator, three answers
A hypothetical chain reports 400,000,000 units of adjusted transfer volume over a year. Its supply breaks down as follows: 120,000,000 units issued in total, of which 20,000,000 sit in locked vesting contracts, leaving 100,000,000 circulating. Of that circulating amount, 30,000,000 are in the staking deposit contract and 20,000,000 have not moved in more than five years, leaving 50,000,000 of free float.
| Denominator | Supply | Velocity (volume / supply) |
|---|---|---|
| Total supply | 120,000,000 | 3.33 |
| Circulating supply | 100,000,000 | 4.00 |
| Free-float supply | 50,000,000 | 8.00 |
The numerator never changed. The highest of the three answers is 2.4 times the lowest. A report citing velocity of 8.00 and one citing 3.33 can both be accurate descriptions of the identical chain in the identical year, and a reader cannot reconcile them without the supply definition.
Part two: velocity doubles with no change in behaviour
Now hold behaviour fixed. In year two, adjusted transfer volume is again 400,000,000 units and total and circulating supply are unchanged. The only difference is that staking grew from 30,000,000 to 50,000,000 units and the dormant cohort grew from 20,000,000 to 25,000,000 units.
| Measure | Year 1 | Year 2 | Change |
|---|---|---|---|
| Adjusted transfer volume | 400,000,000 | 400,000,000 | none |
| Staked supply | 30,000,000 | 50,000,000 | +20,000,000 |
| Dormant over five years | 20,000,000 | 25,000,000 | +5,000,000 |
| Free-float supply | 50,000,000 | 25,000,000 | halved |
| Velocity against circulating supply | 4.00 | 4.00 | none |
| Velocity against free float | 8.00 | 16.00 | doubled |
Free-float velocity doubled from 8.00 to 16.00 with an identical numerator. Anyone reading that series as a behavioural signal would report a dramatic increase in usage in a year when the total transferred was, by construction, exactly the same. The circulating-supply view has the opposite failure: it reports no change at all, concealing that the pool of coins actually able to move was cut in half.
Part three: why two chains cannot be compared
Two hypothetical chains, each with 100,000,000 units circulating and 200,000,000 units of adjusted transfer volume in a year. Chain A has no staking mechanism, so its free float equals its circulating supply. Chain B has 50,000,000 units staked.
| Measure | Chain A | Chain B |
|---|---|---|
| Circulating supply | 100,000,000 | 100,000,000 |
| Staked supply | 0 | 50,000,000 |
| Free-float supply | 100,000,000 | 50,000,000 |
| Adjusted transfer volume | 200,000,000 | 200,000,000 |
| Velocity against circulating supply | 2.00 | 2.00 |
| Velocity against free float | 2.00 | 4.00 |
On the circulating measure the two chains look identical. On the free-float measure, the coins that can actually move on Chain B turn over twice as fast. A cross-chain velocity table built on circulating supply systematically flatters chains with large staked shares.
What Can Make the Interpretation Wrong?
- No stated supply definition. The most common failure. Three defensible denominators produce three answers, and without the label the figure carries no information.
- Attributing a denominator move to behaviour. A rise in staking or a large vesting unlock moves velocity without anyone transacting differently.
- Using raw transfer volume as the numerator. Change outputs and internal exchange movement inflate it, so the velocity inherits an inflation factor that varies over time.
- Treating circulating supply as objective. It is an editorial judgment about which allocations are locked, and publishers reach different totals for the same asset.
- Ignoring the dormancy threshold. Free float depends on where the line between dormant and active is drawn, and moving that line moves the whole series.
- Comparing chains with different lock-up structures. A chain with a large staked share posts a lower free float and a higher free-float velocity for the same activity.
- Naive annualisation. Multiplying a daily figure by 365 assumes the sampled day is typical, which is a strong claim for an event-driven series.
- Reading low velocity as bullish or high velocity as bearish. Neither direction has a stable relationship with anything, and the same move can arise from four different underlying situations.
Cross-Network and Provider Comparison
The largest structural divide is consensus. A proof-of-stake chain makes a large share of supply immobile at the protocol level, with an on-chain deposit contract making the amount directly observable. A proof-of-work chain has no such mechanism, so its immobile share consists only of dormancy and loss, both estimated from output age rather than read off the ledger. Comparing free-float velocity across the two compares a measured quantity with an estimated one.
Ledger model matters too. On a UTXO chain every output carries the height at which it was created, so the age distribution of supply is directly computable. On an account chain a balance has no age of its own, only the account last-activity timestamp, and a small incoming deposit resets that timestamp for the entire balance. The same dormancy threshold selects a different kind of thing on each model.
Token design is the third divide. A token with a multi-year vesting schedule has a circulating supply that steps upward at cliff dates, so its velocity steps downward at those dates regardless of activity. Provider differences on a single asset come mostly from the circulating-supply judgment and the dormancy threshold; on reconciling such disagreements see the Swoopr Investment guide to on-chain data quality methodology.
Advanced Analytical Methods
Decompose the change into numerator and denominator contributions
Because velocity is a ratio, its log change is the log change of transfer volume minus the log change of supply. Publishing those two contributions beside the velocity change makes the artifact case impossible to miss: a velocity move with a zero numerator contribution is entirely a supply story.
Supply-age band velocity
Instead of one free-float threshold, compute velocity against several age bands of supply: units last moved within 30 days, within a year, within five years. The resulting family of series shows which part of supply is doing the moving, and removes the need to defend one arbitrary cutoff.
Staked share as a companion series
Always publish the staked share of supply next to free-float velocity. Seeing them together prevents a staking-driven denominator contraction being read as a usage expansion.
Consistent supply across velocity and valuation
Whichever supply definition is chosen must also be used in supply-based valuation: a portfolio using free float for velocity and circulating supply for market capitalisation is internally inconsistent. See Coin Metrics: Valuation Metrics for the valuation side.
Velocity Definition Comparator
Recommended tool: a comparator that computes velocity against total, circulating, and free-float supply simultaneously from one shared numerator, and reports the spread between them as a first-class output.
Its inputs are the transfer-volume series and whether it is raw or adjusted, the interval, the staked and locked supply series, and the dormancy threshold used for free float. Its outputs are the three velocities, the supply components stacked so the reader sees what was subtracted, and a decomposition of any velocity change into numerator and denominator contributions.
Two design constraints follow from this page. It must refuse to display a velocity without its supply definition attached, and it must show the numerator series alongside the ratio, because the artifact case is invisible in the ratio alone.
Practical Checklist
- I know which supply definition the denominator uses.
- I confirmed the numerator is an adjusted transfer series.
- I plotted numerator and denominator separately, not only the ratio.
- I noted the spread across all three supply definitions.
- I checked the staked share of supply over the same window.
- I checked for vesting cliffs or unlock events inside the window.
- I know the dormancy threshold used for free float.
- I did not annualise a single day of activity.
- For any cross-chain comparison, I checked the lock-up structures.
Frequently Asked Questions
Why can velocity change when nobody transacted differently?
Because velocity is a ratio and the denominator moves on its own. If coins flow into a staking contract or a cohort crosses a dormancy threshold, free float falls and velocity rises with an unchanged numerator. In a worked hypothetical where transfer volume stayed at 400,000,000 units while free float fell from 50,000,000 to 25,000,000, free-float velocity doubled from 8.00 to 16.00 with no behavioural change at all.
Can token velocity be compared across two blockchains?
Only after aligning the supply definition, and often not even then. Two chains with identical circulating supply and identical transfer volume post identical circulating velocity, but if one has half its supply staked its free-float velocity is twice the other. A cross-chain velocity table built on circulating supply systematically flatters chains with large staked shares.
Does low velocity mean people are holding rather than spending?
Only when the denominator is stable and the numerator did the moving. Velocity falls just as easily because circulating supply stepped up at a vesting cliff or because a provider switched supply definitions. Establish which of the two terms moved before making any behavioural claim.
How are lost coins handled in the denominator?
They cannot be identified directly, so free-float estimates use dormancy as a proxy, treating supply untouched beyond some threshold as immobile. That threshold is a choice and it does not separate a lost coin from a patient one. Publishing velocity against several age bands of supply avoids having to defend one arbitrary line.
Should velocity be computed from raw or adjusted transfer volume?
Adjusted, in almost every case, and the choice changes the answer by a large factor. Raw transfer volume includes change outputs, internal transfers between an entity's own addresses and automated movements that do not represent anyone spending anything. Feeding that into a velocity calculation produces a figure describing ledger mechanics rather than economic circulation. Adjusted volume attempts to strip those out, at the cost of depending on the adjustment heuristics, which is why the volume variant used belongs in the definition.
Why is velocity sensitive to the period it is annualised over?
Because it divides a flow measured over a window by a stock measured at a point, then scales the result to a year. A short window captures whatever happened to occur in it and multiplies that up, so a single large day can dominate an annualised figure computed from a week. A long window smooths but blends different conditions together. Two published velocity figures for the same asset frequently differ mainly because one annualised a month and the other a quarter.
Does a token used as collateral have low velocity by design?
Yes, and that is the clearest case where low velocity carries no negative implication. A token locked in a lending protocol, staked to secure a network or posted as collateral is deliberately immobile, so it contributes to the denominator while generating no transfers. A design that requires locking therefore produces low velocity as an intended outcome. Reading that as weak usage inverts the meaning, which is why velocity has to be interpreted against what the token is for.
How does exchange trading affect on-chain velocity?
It suppresses it, because trades executed inside a venue never touch the chain. A token whose activity is concentrated on exchanges shows high economic turnover and low on-chain velocity, while one used mainly for on-chain settlement shows the reverse at the same level of real activity. The share of trading that occurs off chain differs enormously between assets and changes over time, which makes velocity comparisons across assets particularly unreliable.
Is velocity a cause or a consequence of a token's price?
It is an observation, and treating it as a cause is where the framework is most often misused. The identity relating money supply, velocity, price and transaction volume is an accounting relationship that holds by construction, so it can be rearranged to appear to explain price without adding any information. Nothing in the identity establishes direction, and velocity is computed from the same transaction data that reflects market conditions. It describes circulation and does not predict it.
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.
- Coin Metrics: Network Data Glossary: Cross-network supply and ledger definitions underlying the denominator choice.
- Coin Metrics: Transaction Metrics: Adjusted transfer-value definitions used as the numerator.
- Coin Metrics: Valuation Metrics: Supply-based valuation measures that must use the same supply definition as velocity.
- Ethereum.org: Proof of Stake: Deposit, activation, and exit mechanics that make staked supply immobile.
- Bitcoin Developer Guide: Transactions: Output structure that makes coin age, and therefore dormancy, directly computable on a UTXO chain.
- Glassnode: Metric Catalog: A second published catalogue for checking how supply and velocity series are defined.