Token velocity

Token velocity is the rate at which a token turns over – the number of times an average unit is transacted within a period. It matters in token design because under the classical equation of exchange velocity sits in the denominator of a token's implied value: for a fixed volume of economic activity, the faster a token circulates, the fewer tokens the network needs to hold at any instant, and the lower the value each token must carry.

The concept was imported into crypto-economics in 2017 by Kyle Samani, Vitalik Buterin, and Chris Burniske, and it reframed a common failure mode: many "utility tokens" designed purely as a medium of exchange were structurally unable to hold value, no matter how much the underlying network was used.

The equation of exchange

The starting point is Irving Fisher's equation of exchange, MV = PQ: the money supply M times its velocity V equals the price level P times the real quantity of transactions Q. Rearranged, the aggregate value a monetary asset must support is M = PQ / V.

Treat a token as the money of its own micro-economy and the same identity applies. PQ is the annual value of economic activity the token settles; V is how many times each token is used to do so; and M – the market value the network can sustain – is that activity divided by velocity. Two networks settling the same volume of real activity can be worth wildly different amounts purely because tokens in one sit still and tokens in the other are dumped the instant they are received.

The velocity problem

The problem, as Samani framed it, is that a token used only as a medium of exchange has no reason to be held. A user acquires it to pay for a service and the recipient sells it immediately for a currency they would rather keep. Velocity spikes toward infinity, and by M = PQ / V the sustainable market value collapses toward zero – even if the network itself is thriving. Usage and token value decouple.

Buterin made the same point sharply: a token whose only role is to be spent, with no friction against instant re-sale, converges on a value determined by the brief window it must be held to transact – often seconds. This is why "the network is used a lot, therefore the token will be worth a lot" is a non-sequitur. High turnover is not a sign of health for a token's price; it is the opposite.

Velocity sinks: giving people a reason to hold

The design response is to build velocity sinks – mechanisms that give holders a reason to hold the token rather than pass it straight on, lowering effective velocity and letting value accrue. The recurring families of sink are:

• Staking and security bonds – locking tokens to earn yield or to be eligible to provide a service removes supply from circulation.
• Governance rights – where influence is a function of tokens locked, or of a participation record that only a live balance makes usable, holding is itself the way to use the token (see vote-escrow tokenomics, and curation markets, where staked tokens signal and curate).
• Cash-flow or fee capture – a claim on protocol revenue turns the token into a productive asset worth holding rather than flipping (token valuation treats this cash-flow analog directly).
• Redemption / burn-and-mint – a right to redeem the token for something of value at a floor gives holders a reason not to sell below it.

Each sink is a deliberate source of holding demand. Good token design is, in large part, the art of building credible sinks without simply taxing users.

The sinks in production

The 2017 argument was theoretical. The sink that went on to absorb the most supply in practice is vote-escrow, shipped by Curve in August 2020: lock CRV for up to four years and receive non-transferable veCRV, whose weight decays linearly to zero at expiry, so keeping influence means keeping the lock (Curve DAO documentation).

It is measurable, which most velocity claims are not. Read on Ethereum mainnet at block 25,749,903 (2026-08-14 01:10:59 UTC), the vote-escrow contract 0x5f3b5DfEb7B28CDbD7FAba78963EE202a494e2A2 held 850.8m CRV against a total supply of 2.410bn – 35.3% of every token in existence, immobilised by choice. The same read prices the average lock: veCRV weight is locked CRV scaled by remaining lock over the four-year maximum, and veCRV's supply of 780.8m against 850.8m CRV locked implies an amount-weighted average of 3.67 years still to run, 92% of the maximum. Lockers are not parking tokens for a quarter; they are max-locking and re-upping.

That is what a working velocity sink looks like from the outside, and it also shows what the sink costs. Weight that decays unless renewed produced a rental market for it – vote markets, where the influence the sink created is sold by the epoch – so the lock suppressed velocity in the token and re-created it one layer up, in the votes. The burn and fee-switch route (see fee switches and value accrual) takes the opposite trade: no lock, no rental market, and no holding requirement either.

Criticisms and nuance

Velocity is a useful lens, not a law. Several caveats matter. First, velocity is endogenous: it is an outcome of a token's design and its holders' expectations, not an independent input you can set – a strong sink and a credible growth story both depress it. Second, the naive M = PQ / V model treats a token purely as a spending medium and ignores its role as a speculative store of value, which is where much crypto demand actually sits; Burniske's discounted-utility models try to bridge that by discounting future utility value into the present. Third, measured on-chain velocity is noisy – exchange wallets, wrapping, and internal transfers inflate turnover that is not real economic circulation.

The durable takeaway is narrower than "low velocity is always good": a token needs at least one credible reason to be held that scales with the network's success. Without it, adoption and value stay decoupled.

Objections to the velocity argument

Economists and fund analysts objected to the velocity argument within a year of its publication, mostly on the grounds that the equation of exchange was being misapplied. Warren Weber, a former senior research officer at the Federal Reserve Bank of Minneapolis, writing for CoinFund in February 2018, pointed out that in a token economy P is the price of the network's service in tokens, not the dollar price of the token, and V counts how often a token is spent on that service, not how often it is traded for dollars. Rearranging the identity to produce a dollar price per token therefore puts different units on its two sides. The rearrangement works only if the network prices its service in dollars, which leaves open what the token is for.

Anshuman Mehta and Brian Koralewski of the hedge fund Austere Capital called the method circular. A model that forecasts a network's revenue in dollars, then divides by supply and velocity to reach a target token price, has already assumed a price: the network earned that revenue from users who bought tokens at some dollar exchange rate. They and Alex Evans, writing in BlockChannel in January 2018, also objected to how such models treat V. The models choose a value and feed it in, whereas the identity defines velocity only as whatever balances the other three terms, so a chosen V absorbs the error in every other estimate.

Evans kept the core claim and tried to make velocity something a model could predict. The question that matters, he argued, is how velocity moves with the network's transaction volume: if the two rise together, token value stops tracking usage. He modelled holdings with the Baumol–Tobin model from monetary economics, in which a user holds the token balance that minimises two costs, converting into the token and forgoing the return on a store-of-value asset. In his worked example, a hypothetical electricity-payment token called VOLT, falling conversion costs let velocity grow faster than spending on the network, and the token's utility value peaked and then fell while spending kept rising. He also warned against staking and burn mechanisms built to suppress velocity rather than to improve governance or the user's experience.

Scott Locklin, in a December 2018 paper for Brave Software, the company behind the Basic Attention Token, rejected the velocity problem altogether. He argued that the inverse of average velocity is not average holding time, because holding times are skewed: when most tokens sit for years and a few change hands constantly, the average velocity says little about how long a typical token is held. In his worked example of an island economy, a faster exchange does not raise velocity, which rises only when the economy the money serves grows, and locking up or burning most of the supply raises the velocity of the tokens that remain and their price together.

Cathy Barrera of the MIT Cryptoeconomics Lab drew a design constraint from the Catalini and Gans model of token sales covered under token valuation. Holders spend a token when they expect its value to fall and save it when they expect it to rise, so a token built to appreciate leaves its platform idle, and a token that users readily spend is a poor investment. A founding team, she concluded, has to choose which of the two it is building; she named venture funding, or selling tokens to the platform's users rather than to speculators, as ways to fund a platform without the conflict.

How Caper approaches this

A caper does not ask its token to be a medium of exchange, and it does not use a lock. Its sink is the shape of the redemption claim. A member's exit share follows their earned vote weight, not the size of their holding alone. Vote tokens are minted by voting and by nothing else – earned by casting ranked ballots, and never by trading – and they can never be transferred, so the record is a non-tradeable account of the decisions a holder actually turned up for, and it cannot be acquired from anyone who already holds it.

Because the two terms multiply, neither substitutes for the other, and that is where the velocity argument bites. Selling the position drives t to zero and the claim goes with it, however much v has accumulated; conversely a holder who has never cast a ballot has no v at all, however the tokens arrived and however many of them there are, so that bag redeems nothing from the treasury. exit() requires a positive amount of each and aborts otherwise, and the vote tokens it burns must come out of the exiting member’s own account, so a claim cannot be assembled from someone else’s v. Exiting burns the vote tokens presented, so the claim cannot outlive the position that earned it. See leaving a caper for the full mechanics.

The complication this design used to carry – that v accrued to turnover, the very quantity the velocity literature treats as the problem – was removed at the 11 September 2026 redeploy, which stopped trades minting v altogether. Turnover now earns nothing at all. What keeps v a sink rather than a subsidy is that it is worth nothing on its own: it cannot be sold, and it pays only in combination with a position still held at the moment of exit. Casting a ballot earns the right to redeem; only holding lets it be exercised.

The bonding curve reinforces this: buys and sells settle against a reserve rather than a thin order book, so the token's design centres on a position you hold and can later redeem, not a chip you pass along. The claim is not that Caper solves velocity. It is that the velocity literature says a token needs at least one credible reason to be held, and here the reason is a redemption claim that only a held position can exercise – which is why personal tokens are coherent as a held stake rather than a currency.

References

• Kyle Samani, “Understanding Token Velocity,” Multicoin Capital (2017)
• Vitalik Buterin, “On Medium-of-Exchange Token Valuations” (2017)
• Chris Burniske, “Cryptoasset Valuations” (2017)
• Equation of exchange (MV = PQ) – the classical Fisher identity
• Curve DAO: Vote-Escrowed CRV (veCRV) – the production velocity sink, documentation
• Warren Weber, “The Quantity Theory of Money for Tokens,” CoinFund (2018), archived copy
• Anshuman Mehta and Brian Koralewski, “MV=P…Que? Love and Circularity in the Time of Crypto,” Austere Capital (2018)
• Alex Evans, “On Value, Velocity and Monetary Theory: A New Approach to Cryptoasset Valuations,” BlockChannel (2018), archived copy
• Scott Locklin, “Token Economics: Considering ‘Token Velocity’,” Brave Software (2018)
• Cathy Barrera, “Tokens: Investment Vehicle or Medium of Exchange (Not Both),” MIT Cryptoeconomics Lab (2018)