Every ranking site presents the same handful of numbers, and each of them is a calculation rather than an observation. That would be fine if the calculations were stated, but they are not, and the gap between what the number is called and what it contains is where most confident wrong conclusions come from.
Market capitalisation is the last traded price multiplied by the reported circulating supply. It is arithmetic performed on two other numbers, not a quantity anybody measured, and it implies something false in a specific and consequential way: it suggests that amount of money is present in the asset, when the only money present is what participants are currently willing to trade with.
The distinction becomes obvious when you ask what would happen if everybody tried to sell. The answer is not that the market capitalisation would be distributed among them, it is that the price would fall until the available bids were exhausted, and the total received would be a small fraction of the headline figure. That is not a criticism of crypto specifically, the same is true of any equity, but the ratio between the two numbers is far more extreme here because the depth supporting a given capitalisation is much thinner.
What replaces it, when you need a size, is the depth of the market: how much could actually be bought or sold within a tolerable distance of the current price. That number is computable from an order book, it is orders of magnitude smaller, and it is the one that determines what any position you take is worth on the way out.
Market capitalisation answers what would this be worth if the last price applied to every unit. Nothing about the market guarantees that it would.
The second half of that multiplication is even less solid than the first. Circulating supply is meant to count units available to trade, excluding those locked, unissued, or held by the project itself, and there is no standard governing what belongs in each category. The figure is typically supplied by the project, sometimes adjusted by the data provider, and different providers publish different numbers for the same asset on the same day.
The ambiguity is not academic. Tokens held in a treasury the project can sell at any moment are excluded on some methodologies and included on others, and the difference can change a reported capitalisation by a large multiple. Units in a staking contract that can be withdrawn after a queue are neither clearly locked nor clearly circulating. Coins in wallets whose keys are provably lost are counted as circulating everywhere, despite being the one category that is definitively not.
The practical consequence is that comparing the market capitalisation of two assets is comparing two numbers computed by different people using different rules. Where the comparison matters, the supply figure has to be checked against the issuance schedule and the unlock calendar directly, both of which are public for any serious project and neither of which appears in a ranking.
Reported volume is the sum of what venues say they traded, and there has never been a reliable mechanism for verifying it. The incentive to overstate is direct: rankings drive traffic, and traffic drives revenue, so a venue that reports a larger figure than it executed receives a benefit at no visible cost. Multiple independent studies over the years have concluded that a substantial fraction of reported crypto volume did not correspond to genuine trades between unrelated parties.
The mechanisms are known. Trading against yourself produces volume without transferring risk. Fee structures that pay participants for volume make it profitable to generate it. Aggregators that sum venues without weighting for credibility propagate the inflation into the headline number everybody quotes. None of this requires anybody to lie in a legally exposed way, which is part of why it persists.
What is verifiable is narrower and more useful. Depth in an order book can be observed directly, and it is much harder to fake because it has to be real enough to trade against. The cost of executing a given size, measured yourself, is the ground truth that volume was supposed to be a proxy for. Where you need to compare venues, comparing what your own order costs on each of them replaces the entire question.
Total value locked is meant to describe how much capital is deployed in a protocol, and it double counts by construction. Deposit an asset, receive a token representing that deposit, deposit that token elsewhere, and the same underlying capital has been counted twice. Chains of this kind are common and can be several layers deep, and the aggregate figure sums all of them.
It also moves with price rather than with activity, which makes it a poor measure of adoption. A protocol whose deposits are unchanged in unit terms will report a rising figure whenever the deposited asset appreciates, and a falling one when it declines, and neither movement reflects anything about how many people are using it or what it earns. Reports that describe a rising figure as growth are describing the price of the collateral.
Where the concept is useful is inside a single protocol over time, in unit terms rather than in currency terms, which strips out both problems. Comparing the figure across protocols, or quoting an industry-wide total, combines double counting with a price effect and produces a number whose movements are dominated by neither of the things it claims to measure.
Fully diluted valuation multiplies the current price by the total eventual supply rather than the circulating one, and it exists because the difference between those two numbers is frequently enormous. A project trading with a small fraction of its eventual supply released has a fully diluted figure many times its reported capitalisation, and the units that make up the gap will arrive on a published schedule.
That schedule is the most under-read public document in the sector. It specifies when allocations to the team, to early investors and to reserves become transferable, and those dates are known years in advance. Somebody buying at a moment when a large tranche unlocks in the following weeks is buying into a supply increase whose timing was never a secret, and the price reaction to unlocks is one of the more studied regularities available.
Neither figure is the right one to use alone. The circulating capitalisation describes what is trading now; the fully diluted figure describes what the same price would imply for everything that will exist. The distance between them, together with the calendar, is the actual information, and it takes one look at the project's documentation to obtain.
The unlock calendar is public, dated years ahead, and consulted by almost nobody. It is the closest thing in this sector to a known future supply shock.
Holder counts are quoted as a measure of how widely distributed an asset is, and an address is not a person. One person routinely controls dozens, a single platform address can hold balances for millions of customers, and creating a thousand addresses costs nothing beyond the transaction fees to fund them. The number can be increased deliberately and cheaply, and it frequently has been.
The same problem afflicts active address counts, which are quoted as a measure of usage. Automated activity, internal transfers within a service, and addresses generated per transaction by design all inflate the figure without corresponding to additional users. A rising count is consistent with growing adoption and equally consistent with one participant restructuring how they operate.
What can be read from the ledger, carefully, is concentration rather than distribution: how much of the supply sits in the largest addresses, and whether those addresses were funded from a common source. That is a bounded, checkable observation, and it answers a more useful question than the holder count, which is who has the ability to move the price.
Not everything is compromised, and the survivors are the ones nobody puts in a headline. Order book depth is directly observable, hard to fake because it must be tradeable, and it answers the question every other size metric was pretending to. Realised transfer volume on a public ledger, filtered for internal and automated movement, is verifiable by anyone willing to do the filtering. And protocol revenue, where a protocol has any, is a cash flow that can be traced on-chain and compared with a valuation the way any business would be.
All three share a property that explains their absence from rankings: they are laborious to compute and they cannot be inflated by the entity being measured. That is precisely what makes them worth the effort, and it is why the numbers that circulate most widely are the ones whose calculation is controlled by the parties they describe.
Four questions handle most of it and none require specialist tools. Who computed this number, and do they benefit from it being larger. Is it a measurement of something or an arithmetic result derived from other numbers, and if derived, what are the inputs. Does it move with price, which would make any change ambiguous between a price effect and an activity effect. And is there a directly observable quantity that answers the same question, which there usually is.
Applying that to a ranking page is a short and slightly deflating exercise. Most of what is presented as data is a calculation whose inputs are supplied by interested parties, and the few quantities that are genuinely observed are not on the page. That does not make the sector uniquely dishonest; equity markets took a century to build the reporting standards that make their numbers comparable. It does mean that anybody reasoning from these figures without knowing their construction is reasoning from something other than what they think.
It is useful as a rough ordering of scale and useless as a measure of money present. It is the last price multiplied by a reported supply figure, so it inherits the uncertainty in both. When size matters, the depth of the order book answers what you actually wanted to know and is orders of magnitude smaller.
Because there is no standard for what counts as circulating. Treasury holdings, staked units behind a withdrawal queue, and unvested allocations are categorised differently by different providers, and the figure usually originates with the project itself. Two sites can therefore publish capitalisations that differ by a large multiple for the same asset.
Nobody can state a precise figure, and multiple independent studies over the years have concluded that a substantial fraction does not correspond to genuine trades between unrelated parties. The incentive to overstate is direct, since rankings drive traffic. Order book depth and your own measured execution cost are the verifiable alternatives.
Two things. It double counts by construction, because a deposit receipt token deposited elsewhere counts the same capital twice, and chains of this can run several layers deep. And it moves with the price of the collateral rather than with activity, so a rising figure can mean nothing happened except that an asset appreciated.
Look at both, and at the calendar between them. The circulating figure describes what trades now, the diluted figure describes what the same price implies for everything that will exist, and the unlock schedule says when the gap closes. That schedule is public, dated years ahead, and read by almost nobody.
Not reliably. An address is not a person: one individual can control many, a single platform address can hold balances for millions of customers, and generating addresses is cheap. Concentration in the largest addresses, and whether they share a funding source, is the bounded observation worth making instead.
The laborious ones. Order book depth, because it has to be real enough to trade against. Realised on-chain transfer volume, once internal and automated movement is filtered out. And protocol revenue where it exists, because it is a traceable cash flow. All three share the property that the entity being measured cannot inflate them.
The pattern is not, the degree is. Equity markets took roughly a century to build the reporting standards that make their figures comparable, and those standards exist because the same incentives caused the same problems. What is specific here is that the most widely circulated numbers are still computed from inputs supplied by the parties they describe.