Count an asset once — leave derivatives and claims aside.
CoinSpectrum is an independent cryptocurrency ranking with a single guiding rule: an asset counts once. It exists to draw a line between what a cryptocurrency actually is and what has been sold under the label for the last decade.
A cryptocurrency has its own base-layer blockchain and its own consensus mechanism. That is the whole test. The ranking counts every chain that meets it — 527 at the time of writing — and lists everything else in its own section, with its peg, its backing and its host chain named plainly. Wrapped coins, stablecoins, staked derivatives, tokenised bonds and tokens are all listed. None are added to the cryptocurrency total, because the thing a claim represents is already counted.
The two framings are one rule. A claim on an asset — a wrapped coin, a stablecoin, a staked derivative — cannot run on its own blockchain: its value is another asset that already has one. So the classification test excludes claims automatically, and "counts once" is what falls out when the test is applied. One framing is the mechanism; the other is what a reader sees.
The rule is technical and neutral. It does not ask whether a chain's founder can freeze accounts, whether a company holds most of the supply, or whether the roadmap includes the phrase "reserved for the team." Those questions matter, but they are answered on each asset's page — not in whether the asset is counted. The ranking is what the rule finds; the coin pages are where the site says what to think of what the rule found.
The last ten years read as a series of trends that each borrowed the word "cryptocurrency" long enough to raise money, collapse, and hand the reputational bill to the chains that had nothing to do with them.
Many Bitcoin forks were not attempts to improve the protocol. They were an airdrop-value extraction mechanic: snapshot every holder, launch a chain with a small rule change, watch every original coin holder receive "free" coins on the new chain, list on exchanges, let the market price them, sell. Roughly two dozen fork-era attempts followed the pattern. Most had no mining population within months and no market within a year. The reputational bill for "yet another Bitcoin fork" landed on chains that had nothing to do with any of them — the counting rule confirms it: what a mining population no longer extends isn't a chain the site can find.
Companies raised money against a whitepaper, a website and a list of "Advisors" — well-known industry figures whose names appeared to buy credibility, often without their meaningful involvement. Announced partnerships that did not exist. Tokens sold that were not cryptocurrencies by any honest reading of the word, but were counted as such for every purpose the market cared about. Most of what was raised is gone.
Governance tokens issued to bootstrap protocols that never earned the yield they promised. JPEGs marketed as investable assets. Move-to-earn apps whose supply schedules guaranteed the tokens would trend to zero. Rug-pulls, exit-scams, financial pyramids in blockchain wrappers. Each cycle attracted its own generation of newcomers and handed each one a loss they attributed to "crypto" rather than to the specific mechanism that had failed.
Terra's UST drained to zero in a week, taking LUNA with it. The failure had nothing to do with any base-layer chain's consensus mechanism and everything to do with an algorithmic peg that was never going to survive sustained redemption pressure. The word "cryptocurrency" absorbed the damage anyway.
Every one of these was a distinct thing, with a distinct failure mode, and none of them was Bitcoin — or anything with Bitcoin's properties. Lumping them together under one word made every collapse feel like a collapse of the idea itself. It wasn't. The idea kept running, block by block, the whole time.
One failure mode isn't an era. Mt Gox in 2014. Cryptopia, QuadrigaCX and CoinExchange.io in 2019. Mercatox soon after. Celsius, Voyager, FTX and BlockFi in 2022. Bittrex Global in 2023. Dozens more between and around them, from small operators that quietly halted withdrawals to large ones that filed for bankruptcy in the same week. Companies that took custody of user coins and treated the balance sheet as fungible with them until they couldn't. None was a protocol failure — the cryptocurrencies in this ranking kept producing blocks the whole time — but each collapse dragged the word "cryptocurrency" with it in the coverage that followed. Custodial exchange failure isn't a phase of the industry; it's the baseline hazard of any custodial arrangement, and it ran underneath every era above. What custodial exchanges do differently from non-custodial ones, and where each design falls short, is a longer piece on its own.
They are not tokens on someone else's chain. They are not ICOs. They are not premined supplies distributed to a board. They are not JPEGs marketed as investable. They are not governance tokens for protocols yet to be built. They do not have "Advisors" pages.
Not every chain in the ranking clears every bar equally — some have significant company or founder concentration, and the site says so on their coin pages. But every one of them has its own blockchain and its own consensus mechanism, which is more than the trends above could ever say.
Count what is one asset once. List everything else as what it is. Say clearly, in one place, what the rule is; and say clearly, on each page, what we think of what the rule found.
Every category the market puts under the word "crypto." What each one actually is, and how it is treated on the site.
An asset with its own base-layer blockchain and its own consensus mechanism. Something that mints itself on a rule no committee can override, and settles peer-to-peer without an intermediary who could freeze it. Four properties fall out of that design — no mint, no confiscation surface, pseudonymity, own consensus on its own chain — and an asset carries all four or it is something else. The long explanation walks through the etymology and the whitepaper.
A ledger entry on someone else's chain. An ERC-20 rents Ethereum's consensus; a BEP-20 rents BNB Chain's; an SPL rents Solana's. The token can't produce blocks — it has no chain — and its rules can be changed by whoever holds the deployer keys. The site lists tokens with their host chain named, and never counts them as cryptocurrencies.
A token pegged to a real-world asset — usually a US dollar, sometimes a euro, gold, or another currency. The issuer holds the mint (they decide when more enters supply) and holds a freeze primitive (they can and do blacklist addresses at law-enforcement request). Two of the four whitepaper properties are inverted. What a stablecoin does — a dollar on crypto rails — is a real product; it is not a cryptocurrency in the definitional sense.
A token issued on one chain that represents an asset held on another. Wrapped BTC on Ethereum is a claim on a Bitcoin held in custody somewhere. cbBTC is a Coinbase-custodied Bitcoin re-issued as an Ethereum token. Both trade at a price tied to the underlying by design. Both are listed on the site; neither is counted as a cryptocurrency, because the Bitcoin they represent already is.
A tradable receipt for a coin locked in staking. Lido's stETH is a receipt for staked Ether; Rocket Pool's rETH is another. The receipt earns the staking yield while it moves on-chain like an ordinary token. The underlying Ether is already counted; the receipt is listed as a claim on it.
A smart-contract vault into which a user deposits one asset and mints another against it, with a fixed collateralization ratio. MakerDAO's DAI is the reference case: deposit ETH, mint DAI worth up to two-thirds of the ETH's value; if ETH falls too far, the vault auto-liquidates. CDP-issued stablecoins are non-custodial by design — no issuer holds the keys — which puts them in a different category from Tether or USDC, though the site still classifies them as stablecoins because their intended peg is a dollar.
A token whose value proposition is a cultural reference. Dogecoin, Shiba Inu, Pepe, and thousands of newer ones. Some are premined to a founder's wallet on day one; some distribute through fair launches. What none of them do is carry a claim to solve a technical problem — the appeal is the meme itself. The site lists memecoins in a separate category; the largest ones by market cap are shown, the long tail is gated by volume.
A non-fungible token: a chain entry that says who owns a specific item — a JPEG, a domain, a game asset — where each item is unique rather than one-of-many. The chain entry says who owns the URL, not what the URL points to; the underlying artefact usually lives on someone's server and can go dark. The site does not track NFT collections. They are not currencies in any sense the word normally carries.
Both count every wrapped coin, staked derivative and stablecoin as if it were a distinct cryptocurrency. Bitcoin gets counted; Wrapped Bitcoin gets counted again; a Coinbase-wrapped BTC gets counted a third time. The number that comes out isn't "how many cryptocurrencies exist" — it is a list of every ticker on every venue. This site says one asset counts once.
A fork is counted only if it kept its own mining population, its own market, and its own chain producing blocks. BTC, BCH and BSV meet the test and are counted separately — three cryptocurrencies. The two dozen fork-era attempts that lost their mining and market within a year are not counted, because the rule cannot find them: the chains stopped.