Every asset the site lists carries a grade from A++ to F, built from criteria anyone can check: twenty for a coin and ten for a token. It states the site's position in numbers: a cryptocurrency with its own blockchain, minted openly by whoever does the work, with no premine, no insiders' sale and a deep market is the standard the industry measures itself by. Tokens issued and controlled by a group carry more risk, and memecoins most of all. The grade rates how an asset is built and how deep its market is. It is not a price forecast and not investment advice.
Each criterion is scored from 0 to 10, and all are weighted equally: the score out of 100 is their average times ten.
| Criterion | How it is scored |
|---|---|
| Consensus: leader ladder | Consensus families ranked by the market cap of their largest coin, recomputed with every build, with the token classes after every family on the same ladder. The places are spaced evenly from 10 at the top to 0 at the bottom, to one decimal. A token takes its asset class instead, placed below every coin family (see tokens below). |
| Consensus: family ladder | Consensus families ranked by the combined market cap of all their coins, on the same scale. Two criteria rather than one average, because the two can disagree: one family can hold the single most valuable coin while another holds the most value overall. |
| Sound Money Qualities (6 checks) | The first section of the Cryptocurrency Standards: Bitcoin was built as sound money, its announcement naming the trust conventional money needs, that “the central bank must be trusted not to debase the currency” (Satoshi Nakamoto, P2P Foundation, February 2009). Six qualities of sound money, 1 point each, scored out of 10 on what is measured, at least four; the range counts the rest as failed and as passed. They read facts other criteria also score, by the operator's choice: one criterion, the overlap accepted. Scarcity: half the supply scale (see supply: 21 million scores 10), half a fixed cap that holds, 10 with one and 0 without (a fixed maximum that no issuer, key or on-chain vote can mint past without node operators choosing new software; the basic list's limited-supply check until method version 30) · durability: five full years or more with no halt, freeze or reversal in the record · fungibility, by the standard definition, units completely interchangeable with any other of the same kind and value: no coin or address the protocol can freeze, blacklist or set apart (the same fact censorship resistance reads) · divisibility: the smallest unit the ledger records worth a cent or less at today's price · portability: fees of a cent or less and capacity for real demand, 5 or more on the throughput scale; high fees and low throughput both make a coin harder to move · verifiability: open-source node software and a way to check the ledger without trusting a server, light clients or pruned nodes. Coins only. |
| Privacy (5 checks) | Five protocol properties, the share passed out of 10, at least two measured: whitepaper privacy, keys not tied to an identity, the paper's Section 10 ("keeping public keys anonymous"): a chain whose own rules make an identity check a condition of holding or sending the coin fails, while exchange or wallet checks do not count · a UTXO ledger, the paper's Sections 2 and 9 and its "new key pair should be used for each transaction": a wallet takes a new key for each payment, where an account on an account-based chain gathers every coin its holder owns and links every payment to it (the basic list's check until method version 30) · network-level privacy, the node hiding where a transaction came from (Dandelion++, or Tor or I2P transport, built in and on by default; Tor a user must set up does not count) · hidden amounts (confidential transactions, RingCT, shielded pools, Mimblewimble) · hidden sender and receiver (ring signatures or zero-knowledge proofs, and one-time or stealth addresses). Optional features count where any user can use them on mainnet; wallet tools such as CoinJoin, sidechains and layer 2s do not. Coins only; a token has no protocol of its own. |
| Scalability (fees, throughput, block time, block space, features) | How far the chain can grow in use without growing expensive or heavy to follow: the average of five parts, to one decimal: four measures on a log scale and three features. Fees: the typical fee of a simple transfer, $0.001 or less 10, 3.3 points off for each tenfold step, $1 or more 0 ($0.01 6.7 · $0.10 3.3); a chain that is expensive to use is one fewer people use, whatever it can carry. Throughput: capacity in simple transactions a second, the limit set by block size and interval or, without a fixed limit, the sustained rate shown on mainnet, against real demand: the ten layer-1 networks carrying the most real transactions a second, vote and other consensus transactions left out (measured 2026-10-28). The highest single-day average among them scores 10, 2,478 a second (Solana, 2026-08-19); the median of their 30-day averages 5, 119 a second; on a log scale, the same step for each tenfold below. A chain that can carry what the busiest network carried on its busiest day scores full marks, whatever more it could carry. Until method version 30 the scale ran from 1 a second (0) to 1,000 (10). Block time: ten minutes 10, falling away either side by 3.6 points for each tenfold step, to 0 at one second (one minute 6.4 · ten seconds 2.8 · an hour 7.2). Every block adds a header that each node and lightweight client must fetch and keep, so faster blocks mean more headers, more traffic and more cost to sync, the reason the whitepaper sizes the chain on ten-minute headers (Sections 7 and 8); slower blocks keep payments waiting. Block space: the block size limit per ten minutes (the limit times 600 divided by the block time), so fast blocks of moderate size are not punished for their interval twice, 0.1 MB or less 0, 1 MB 4, 4 MB 6.4, 32 MB or more 10. A chain limited by gas, compute units, mass or a transaction count rather than bytes is converted to bytes from that limit (a gas limit divided by the gas charged for a byte of transaction data, and so on); a chain with no hard limit has no block-space figure, and the other parts average alone. Features, three checks, the share passed out of 10: payments safe to accept within seconds (finality within about five seconds, or unconfirmed payments the network protects: no replace-by-fee, double-spend proofs or instant locks; fast blocks alone do not count) · pruned nodes, a full node that validates the chain and drops old history · light clients that verify payments from block headers or proofs without trusting a server. Payment channels, a second layer, were a feature until method version 30: the grade scores the chain itself. At least two parts, one of them fees or throughput, or no score. Coins only. |
| Security (miner pay and 7 checks) | Nine points. Miner pay, two of them: what miners earn a day, issuance and fees, on a log scale, $1,000 or less 0, 2.5 points for each tenfold step, $10 million or more 10 ($10,000 2.5 · $100,000 5 · $1 million 7.5), counted as two of the nine points; it is what an attack must outspend, and a chain that is not mined scores 0. Then seven checks, 1 point each: its own hashrate, at least half the largest on its mining algorithm and not merge-mined (its blocks found by another chain's miners as a by-product) · largest pool or validator under 50% · largest three under 75% · more than one maintained client · node software is open source (the reference client published under an open-source licence, so anyone can audit what the network runs; closed source fails) · real on-chain use, 1,000 or more transactions a day · no 51% attack or reorganisation deeper than six blocks of live transactions in five years. Scored out of 9 on what is measured, at least six points' worth. Until method version 28 miner pay was two checks, over $1 million and over $100,000 a day. Tokens, with no network of their own, 0. |
| Programmability (tokens, contracts, 5 checks) | What the protocol lets its users build beyond payments: nine points, scored out of 10 on what is measured. Token issuance, 2 where anyone can issue tokens the chain's own nodes validate (ERC-20, CashTokens, native assets), 1 where tokens exist only through a metaprotocol read by indexers (Ordinals, Runes, Omni), 0 neither. Tokens on another chain do not count: the grade scores the chain itself (a second layer counted 1 until method version 30). Smart contracts, 2 for a general-purpose contract system on the base layer in which applications such as exchanges or lending run on mainnet (EVM, WASM, Move, Plutus, CashScript with introspection), 1 for scripting limited to spending conditions (Bitcoin Script), 0 neither. Then five checks, 1 point each: multisig and time locks, m-of-n and time-locked spending without a third party · atomic swaps with other chains, through hash- and time-locked contracts or adaptor signatures · an exchange built into the protocol (XRP Ledger, Stellar; an exchange deployed as a user contract does not count) · cross-chain messaging the chain verifies itself (IBC, XCM; third-party bridges do not count) · non-fungible tokens the chain's rules carry, natively or through a contract standard its nodes run (inscriptions read by indexers do not count). Protocol level only. No score until smart contracts and at least five points' worth are researched; the range counts the rest as failed and as passed. Token issuance was part of development until method version 29. Coins only. |
| Censorship resistant | No authority that can halt the chain or freeze funds, 10 or 0: the whitepaper's Section 2, a coin's owner the holder of its key, and the chain no permission to halt it, blacklist an address or reverse a payment. A chain whose issuer, foundation or validator council can do any of these fails, as does a federation or narrower set of block producers that can; a mined chain passes unless its record shows otherwise. A check of the basic list until method version 30. |
| No central team | No single company, foundation or lab controls the reference client or the roadmap, 10 or 0: the paper's network of nodes that need little coordination. A chain whose production software one organisation maintains and can push a breaking change to fails. Whether the code is published is scored under security. A check of the basic list until method version 30. |
| Who can mint | Who can produce blocks, and so who issues new coins, as a ladder of consensus families: proof of work first, proof of stake second, hybrid proof of work and stake third (it inherits both models directly, so it follows them whatever its size), then every other family, proof of space included, by the combined market cap of its chains, then Other, a bucket of unclassified models rather than a family, below every known family and below a chain of a known family that one company runs, and last, at 0, tokens a founder or company issues, such as USDT by Tether. The places are spaced evenly from 10 to 0, to one decimal, and recomputed every build (the ladder today is below). A family is how a chain makes its blocks; BFT, a way blocks become final, and DAG, a shape of ledger, run inside several families and are not families of their own. In a family that mixes open and permissioned variants (proof of stake, hybrid, Other), a chain whose block producers are admitted from a permissioned list loses 1 point; a federated or proof-of-authority chain does not, since admission is what those families are; proof of authority, blocks signed by pre-approved authorities or one company's validators, is its own family. Tokens are not scored on it: they already rank below every consensus, and who can mint more of a token is its own criterion. |
| Nodes | How many machines keep a full copy and check the rules: reachable full nodes from a crawler, or the active validators or block producers where no crawler counts the network (each asset page says which). For chains that speak Bitcoin's peer-to-peer protocol and that no outside crawler counts, the site runs its own crawler: from the chain's DNS seeds it asks every node for the addresses it knows and follows them, counting the distinct IP addresses that complete the handshake on the chain's own network bytes at a block height within 0.5% of the chain's tip (and, where chains share network bytes, run that chain's software). Nodes reachable only over Tor are not counted. A logarithmic scale over the whole range, because going from 1 node to 10 matters as much as going from 100 to 1,000: one node, a single point of failure, scores 0 and 1,000 or more score 10, each doubling worth about 1 point: 2 nodes 1, 4 nodes 2, 10 nodes 3.3, 30 nodes 4.9, 100 nodes 6.7, 300 nodes 8.3, 560 nodes 9.2, to one decimal. Collateral slots and validator keys are not nodes, as one operator can hold any number: a registry of them counts machines by distinct IP address where it lists them, and is otherwise left unmeasured. Not yet counted: left out of the average. |
| Holder concentration | Share of supply held by the 100 largest non-exchange addresses: under 10% scores 10, over 90% scores 0. Being collected. |
| Premine (allocated at launch) | What was allocated at launch: 10 × (1 − the share of today’s supply that founders, a company, a foundation or investors received at genesis or in a documented instamine, without mining it). Developers who mined like everyone else do not count, and neither do reward pools paid out to open miners or stakers. What insiders receive after launch is the next criterion. Bitcoin 10; Ethereum, about 59%, 4.1; anything issued in full by a group, 0. |
| Fundraising | No sale of the coin 10. A sale by its issuer or insiders, whatever the channel (a public sale, a private, seed or venture round, or a company selling its pre-issued supply on the market), lowers the score by everything they had to sell: what they have sold is in the market, and what they still hold they can sell. 10 × (1 − the larger of the share sold and the pre-issued share). Ethereum, about 59% pre-issued, 4.1; XRP, issued in full to Ripple and its founders, 0, however much is sold so far. The share actually sold is shown on each coin’s page. |
| Developer share (after launch) | What developers, founders, a company, a foundation or a treasury receive after launch without mining it: a share of every block reward (Zcash’s founders’ reward and development funds, eCash’s development share, Dash’s and Cardano’s treasuries) or a one-off mint after launch (the Sia Foundation’s 2021 grant). Scored like premine, 10 × (1 − the share of today’s supply received so far), with today’s rate shown beside it because a running share keeps growing. Coins only: a token has no block reward. |
| Supply & Demand | The average of six measures. Exchanges: distinct exchange companies with a live market (Coinbase and Coinbase International count once), 10 or more scoring 10, down to 1. Daily volume: the median day of trading volume over the last 30 days, in dollars, on a log scale, $10,000 or less 0, 2 points for each tenfold step, $1 billion or more 10: $100K 2 · $1M 4 · $10M 6 · $100M 8. Until method version 27 it was a rank among every listed asset, first place 10, falling evenly to 0 below 200th, which moved with every other asset's trading and scored a coin trading $2 million a day level with one trading $500; the rank is still shown beside it. A stablecoin counts every pair quoted in it (BTC/USDT is Bitcoin volume and Tether volume) except pairs of two stablecoins. An asset other than a stablecoin trading more than its whole market cap a day is left out of the volume score and the rank: those volumes, mostly tokenised-stock tokens reporting billions a day, had taken 80 of the top 200 places. Volume is recomputed every minute from the exchanges the site reads (their 30-day median where they quote the asset, the six-hourly snapshot's 24-hour volume otherwise) and shown live on each asset page; the letter uses the volume read when the grade is built. One day's volume moved the score, and with it letters near a band edge; a month's median day does not. Turnover: the 30-day median of daily volume as a share of market cap, the share of the supply the price is actually set by, which daily volume cannot show because it rewards size. One day is too noisy to grade on: it would change the score almost every other day, the 30-day median about one day in twenty, and a median keeps a day of wash trading or a listing spike from carrying the month. Order books deepen with turnover up to a point: in a snapshot of the books on Binance, OKX, Bybit, Gate and MEXC in October 2026, the dollars resting within 2% of the price rose from 0.01% of market cap at under 0.2% turnover to about 0.3% at 3 to 5%; for coins over $10 million the book reached the depth their size predicts by about 4% and gained little beyond 6%, while above 6% a day of volume grew from 10 to 17 times that book to 36 to 100 times. Where little of the supply trades, each trade moves the price further: over the six months to October 2026, a day's trading of 1% of market cap moved the price about 12% where turnover was under 0.2% a day, and about 1% where it was 2 to 5%. Turnover is not a promise of calm: the same test found more active markets more volatile, because large moves draw trading. 5% a day scores 10; away from it the score falls by the same 5.3 points for each tenfold step either way, to one decimal: 0.1% 1 · 0.3% 3.5 · 1% 6.3 · 2% 7.9 · 3% 8.8 · 4% 9.5 · 5% 10 · 10% 8.4 · 15% 7.5 · 25% 6.3 · 50% 4.7 · 100% 3.1. Above 100% a day the volume and the market cap cannot both be right (a supply figure near zero, or wash trading), and turnover is left out of the average. A stablecoin scores 10 from 5% up with no ceiling: its price does not move with its trading, so heavy turnover is what it is for. Recorded every day and shown live; the days before recording began on 26 October 2026 were seeded from CoinMarketCap's daily volume, which runs at its own level (wash trading included), scaled coin by coin to the level of the exchanges the site reads on the days both cover. Order-book depth: the dollars of bids and asks resting within 2% of the price on the spot books of Binance, OKX, Bybit, Gate, MEXC, Bitget, KuCoin, Coinbase and Kraken, as a share of market cap. Turnover says how much of the supply trades; depth says how much of the market cap the books would actually absorb. Read four times a day; each day's median snapshot, then the median of the last 30 days. Log scale: 0.001% of market cap or less 0, 3.3 points for each tenfold step, 1% or more 10: 0.01% 3.3 · 0.1% 6.7. Where a venue serves fewer levels than reach 2% (Bybit, Kraken, Bitget and KuCoin cut the deepest books short) the figure is what the levels hold, a lower bound, and the row says so. A book counts only where its price is within 2× of the asset's, a native coin before a token of the same symbol. An asset no exchange read lists is not measured rather than scored 0. Paper trading: open interest in perpetual futures on Binance, Bybit, OKX, Bitget, Gate, MEXC and Hyperliquid, in dated futures and options on Binance, Bybit, OKX and Deribit, and margin positions on Bitfinex, as a share of market cap. A future, an option or a margin position lets anyone hold or short the price, often on borrowed money, without holding the coin, so a large open interest is paper supply beside the real one, and the price follows leverage and liquidations rather than spot buyers and sellers. 0% scores 10 and 100% or more 0, on a square-root curve, 10 × (1 − √share): 1% 9 · 2% 8.6 · 5% 7.8 · 10% 6.8 · 25% 5 · 50% 2.9; none listed, 10. Read every ten minutes and shown live. Until method version 28 it counted perpetuals alone, on a ladder from 0.5% (10) to 30% (1). Market cap: the size of the market a holder can sell into, on a log scale, $1 million or less 0, 2.5 points for each tenfold step, $10 billion or more 10: $10M 2.5 · $100M 5 · $1B 7.5. Without it a $2 million coin with one busy exchange could sit level with a $1 billion one. |
| Supply (smaller is better) | How many units of the asset there are, or can be, 0 to 10, scored against 21 million. A fixed maximum supply is scored on that maximum alone; an asset without one is scored on the supply already in circulation (sound money's scarcity checks a cap that holds separately). 21 million, Bitcoin's cap, scores 10: enough units for money, few enough to stay scarce. A scarcer asset is less usable as money, 2 points off for each tenfold below (2.1 million 8, 210,000 6, 21,000 4, 210 or fewer 0). A larger one is less scarce: gently to 21 billion, still within what a world currency can hold with scarcity kept (210 million 8.7, 2.1 billion 7.3, 21 billion 6), then 2 points off for each tenfold (210 billion 4, 2.1 trillion 2, 21 trillion or more 0). Until method version 30 the smaller the better, 21 million or less 10 and 21 billion 5. Coins and tokens. |
| Record (origin, claims, halts) | What the people behind a chain have said and done. 10, less 1 for each kind of documented issue, each sourced on the asset page: launched as a joke, its founders saying it has no purpose · claims by its founders or lead company that contradict its own design · the chain halted, or transactions frozen or reversed, by a central party or coordinated validators in the last five years. Court outcomes are not scored, and a chain split is not a deduction. Not yet reviewed: left out of the average. |
| Age (years since genesis) | How long the chain has run: 1 point for each full year since its genesis block, 10 from ten years. A chain that has survived years of markets, attacks and upgrades has shown something a new one cannot. A chain that split from another and kept its ledger, every balance and block before the split, counts from the parent's genesis, since that history is its own: Bitcoin Cash, Bitcoin SV and eCash from Bitcoin's of January 2009, Ethereum Classic from Ethereum's of July 2015. A chain that copies another's genesis block and starts its own ledger at block 1 counts from block 1: Dingocoin's block 0 is Dogecoin's, its block 1 was mined in April 2021. Where no genesis date is researched yet, the first day with a recorded price stands in, which can only understate the age, and the asset page says so. |
| Development (work, improvements, unfixed issues) | How much work goes into the chain, 0 to 10. Steady work, up to 6: half a point for each of the last twelve complete months with commits to the reference client's repository, counted across every branch, merges left out and commits inherited from a forked upstream (Bitcoin Core, go-ethereum, the Cosmos SDK and others) excluded; where commits are not counted or read lower, the release record: more than two releases in two years 6, fewer 2, abandoned 0. Improvements, 1 point for each kind: a security fix, a privacy improvement, a scaling upgrade or a new capability shipped in the last two years. (Token issuance, scored here until method version 29, is part of programmability.) One point per kind rather than per fix, so a codebase that needs many patches does not outscore one that needs few; a kind not shipped costs nothing. Issues, 1 point off each: a disclosed vulnerability still unfixed in the current release, and an exploit of the chain itself in the last two years, a flaw in its protocol or node software used to steal, mint or reverse funds or halt the chain (not a hack of an exchange, bridge or application). Kinds not yet researched form the range. Coins only. |
Tokens are graded on ten criteria: no developer share, since a token has no block reward, and the host chain in its place — the grade of the chain the token is issued on, out of 10, because a token is only as secure and as hard to censor as its chain (tokens on layer-2 networks, which are not graded chains, have no host score). A token has no consensus of its own, so the consensus criterion takes its asset class, placed on the same ladders strictly below every native coin family, whatever its market cap or trading volume: tokens, stablecoins, wrapped and staked claims, tokenised real-world assets and memecoins, ordered among themselves by their largest asset and their combined market cap (the table above). Minting and supply, in place of the Cryptocurrency Standards, asks who can create more, and whether the supply is capped: a fixed supply with no mint authority scores 10, new tokens only through token-holder governance 6, an issuer, foundation, multisig or admin key 0 (stablecoins and wrapped tokens included). Network security stays at 0: a token has no network of its own. Insider allocation is the share pre-issued at launch: a token minted in full when it was deployed is 100% pre-issued, whoever holds it afterwards; only supply emitted openly over time to participants, such as liquidity-mining rewards, is not. Fundraising falls with the share sold, and supply the deployer seeded into a trading pool for the public to buy is a sale — so a memecoin launched "fairly" through a pool is, by these rules, pre-issued in full and sold.
Memecoins are capped at E, whatever they score: a memecoin's own thesis is that it has no purpose, and launchpad memecoins carry the worst record of all — Solidus Labs found that 98.6% of the more than seven million tokens launched on Pump.fun between January 2024 and March 2025 showed the patterns of a pump-and-dump or a rug pull (Solidus Labs, 2025).
| Grade | Score | Cryptocurrencies | Tokens |
|---|---|---|---|
| A++ | 90–100 | 1 | 0 |
| A+ | 82–89 | 7 | 0 |
| A | 74–81 | 27 | 0 |
| B | 64–73 | 64 | 0 |
| C | 52–63 | 80 | 16 |
| D | 40–51 | 14 | 320 |
| E | 25–39 | 2 | 3490 |
| F | 0–24 | 0 | 1874 |
| Pending | — | 309 | 0 |
Counts from the 28 October 2026 build. The highest-graded cryptocurrencies are listed in full.
| Family | Leader | Leader score | Family market cap | Family score |
|---|---|---|---|---|
| Proof of work | Bitcoin | 10 | $1.68T | 10 |
| Proof of stake | Ethereum | 9.1 | $537.5B | 9.1 |
| Federated | XRP | 8.2 | $93.8B | 8.2 |
| Proof of authority | Bittensor | 7.3 | $3.7B | 7.3 |
| Other | Internet Computer | 6.4 | $2.2B | 6.4 |
| Proof of space | Filecoin | 5.5 | $999M | 5.5 |
| Hybrid | Decred | 4.5 | $682M | 4.5 |
| Stablecoins (token class) | Tether | 3.6 | $311.3B | 3.6 |
| Tokens (token class) | WhiteBIT Coin | 0.9 | $116.6B | 2.7 |
| Wrapped and staked claims (token class) | Lido Staked Ether | 1.8 | $103.9B | 1.8 |
| Tokenised real-world assets (token class) | Figure Heloc | 2.7 | $40.1B | 0.9 |
| Memecoins (token class) | Shiba Inu | 0 | $16.0B | 0 |
| Family | Family market cap | Score |
|---|---|---|
| Proof of work | $1.68T | 10 |
| Proof of stake | $537.5B | 8.6 |
| Hybrid | $682M | 7.1 |
| Federated | $93.8B | 5.7 |
| Proof of authority | $3.7B | 4.3 |
| Proof of space | $999M | 2.9 |
| Other | $2.2B | 1.4 |
| Tokens issued by a founder or company | — | 0 |
Proof of work, proof of stake and hybrid proof of work and stake hold the first three places whatever their size: hybrid chains run both of the first two models directly, so they follow them even with a smaller family. Every other family, proof of space included, is a model of its own and follows by market cap. Other comes last among the chains whatever its size: it is not a family but a bucket of consensus models not yet classified, and a chain whose model is known, even one that a single company runs, ranks above it. In a family that mixes open and permissioned variants, a chain whose block producers are admitted from a permissioned list scores its family's place less 1 point; federated and proof-of-authority chains are permissioned by definition and score their family's place.
Both ladders are recomputed with every build. Families and token classes share each ladder, its places spaced evenly from 10 at the top to 0 at the bottom, to one decimal. Token classes come after every native coin family however large they are: a token has no consensus of its own, so it ranks below the chains that do. Among themselves the classes are ordered the same two ways; a token scores the average of its class's two places. Proof of work leads both: its largest coin, Bitcoin, is the asset the whole market is priced against, and its family holds the most value.
Nothing is estimated. A criterion without data is left out of the average and the grade is marked provisional. Holder concentration is being collected for every asset, so every grade is provisional until it is, equally for all. A checklist item not yet researched is left out of that checklist, and the checklist is scored on the items measured. A cryptocurrency whose launch — its premine and any sale to investors — is not yet researched gets no letter at all, only pending: leaving those two criteria out would rate it above a coin whose premine was checked and counted against it. The same holds for network security: a chain with fewer than six of its nine security checks measured, or a mined chain whose miner pay is unmeasured, is pending, so an abandoned chain cannot score well on its known strengths alone.
Leaving a criterion or a check out moves the score, so every provisional grade is published with
its range: the score if everything unmeasured scored 0, and if it all scored full marks. The
asset page states the range in words and grades.json carries
it as range.
Node counts and holder concentration are imperfect measures, and are shown as figures of their own beside the score, not only folded into it. One operator can run many nodes, and a small proof-of-stake validator set can be more independent than a large one run by a few staking services. The largest addresses are not the largest owners: one exchange holds many addresses, one address can hold coins for thousands of customers, and bridges, vesting contracts and burn addresses distort the count. Labelled exchange, bridge, burn and vesting addresses will be left out.
A chain that mints openly, to whoever spends the work, has no issuer who can print more, freeze balances or sell an allocation into the market. A premine or a sale to insiders puts a large share of supply in a few hands before anyone else can earn it, and venture rounds are typically bought at a fraction of the public price and unlocked on a schedule. Tokens add an issuer on top of a host chain, and launchpad tokens and memecoins have the worst record of all: the large majority show the patterns of rug pulls or pump-and-dump schemes. The site's methodology sets out how assets are classified in the first place.
A chain split is not a mark against a chain. Bitcoin Cash, Bitcoin SV and eCash are branches of Bitcoin that share its history, and each is graded on the same criteria as any other chain. What a branch does after the split — the design it follows, the claims made for it, its record — counts like anything else.
The grade measures how closely an asset matches the standard the market itself has set. Bitcoin defined what a cryptocurrency is — its own chain, its own consensus, coins minted openly to whoever does the work, a supply nobody can change at will — and it is still the asset the whole market is priced against. CoinSpectrum does not invent that standard and does not ask anyone to accept it on trust: it measures every asset against it with published criteria, and every input on an asset page carries its source.
Two criteria rank consensus families by market value: one by the family's most valuable coin, one by the combined value of every coin in the family. They do not claim that proof of work is better engineering than proof of stake. They record where the market has put its money — the clearest available signal of which designs people trust with their savings — and that signal is the industry's own standard, not the site's. Both ladders are recomputed from live data with every build and follow the market wherever it goes: if a proof-of-stake coin overtook Bitcoin, proof of stake would take first place on the leader ladder, and if staking chains together held more value than mined ones, it would lead the family ladder too. There are two ladders rather than one because they can disagree — one family can hold the single most valuable coin while another holds the most value overall — and a coin then scores on both facts.
The two ladders are a counterbalance, not a double count. They agree today because proof of work holds both the most valuable coin and the most value overall. Were another family to hold the single largest coin but less value in total, its coins would score high on one ladder and lower on the other, and the family's standing would settle between the two.
The market's weight is stated openly. Three of a coin's twenty criteria read the market: the two consensus ladders and liquidity, which counts market cap; and miner pay, two of the nine security points, reads what miners are paid, which moves with the coin's price. A rise in price can lift them without any change to the protocol. Privacy, censorship resistance, premine, fundraising and developer share do not move with price at all, and each asset page shows the group averages separately, so a reader can see how much of a letter is market and how much is structure.
The Cryptocurrency Standards start from ten properties drawn from the Satoshi 2008 Bitcoin whitepaper. Every check names the section it is drawn from, so a reader can verify the claim rather than take the site's word for it, and the Cryptocurrency Standards post takes them one by one. Since method version 28 each is scored once, where it belongs (the second column); since version 30 the basic list is gone, its four checks moved to sound money, the first section of the standards, to decentralisation and to privacy. The standard is the document, not any chain's behaviour today, so BTC itself falls short on three of them. The criterion was called the Bitcoin Whitepaper standard until 27 October 2026; the checks are standards for any cryptocurrency that the paper happens to state first.
| Property | Scored in | Where it comes from |
|---|---|---|
| Anyone can mint | Decentralisation: who can mint | Sections 4 and 5 (proof of work, and every node working on it) and the conclusion's nodes that "leave and rejoin the network at will": block production is open to any node willing to spend the work. Capped validator sets, elected producers, allow lists, federations and councils fail; proof of work, permissionless proof of stake, proof of space and hybrids pass. |
| Limited supply | Sound money qualities: scarcity | Section 6 ("Incentive"), "once a predetermined number of coins have entered circulation": issuance by a predictable formula, with a fixed cap that no issuer, key or on-chain vote can mint past without node operators choosing new software. An uncapped protocol with no issuer still passes if there is a predictable rule; a protocol where a committee can raise the cap fails. |
| No pre-issuance or developer mint | Decentralisation: premine, developer share | Section 6: the block reward is how coins first enter circulation, "since there is no central authority to issue them", so the first block is mined for its reward on the same rule as every later block. Any amount allocated to founders, a company, a foundation or investors at genesis fails this check, as does any one-off later mint outside the ongoing schedule. |
| No developer tax | Decentralisation: developer share | Section 6 again, after launch: the new coin goes to the creator of the block, so a share of every block reward that is routed to developers, a foundation or a treasury fails this check. Developers who mine like everyone else count in neither issuance check. |
| Censorship resistant: no authority that can halt the chain or freeze funds | Decentralisation: censorship resistant | Section 2's definition of ownership: a coin's owner is the holder of the private key, and the chain carries no "freeze" permission the designer left in. A chain whose issuer or validator council can halt it, blacklist an address or reverse a transaction fails this check. |
| No central team | Decentralisation: no central team | The paper's "network of nodes" picture: no single company, foundation or lab controls the reference client or the roadmap. Whether the code is published is scored under security. A chain whose production software is maintained by one organisation that can push a breaking change fails. |
| Fees under one cent | Scalability: fees, on a scale | Section 1's ("Introduction") "small casual transactions": the paper names the cost of a mediator ruling disputes as what rules microtransactions out, and gives direct payments as the alternative. Sub-cent on-chain fees mean the whitepaper's small-payment use case is actually possible; on-chain fees above a cent fail this check. BTC fails here. |
| Payments safe to accept within seconds | Scalability: features | Section 11's probability analysis of double-spending is built on an attacker's cost of catching up to the honest chain over time; the paper's whole model is an in-the-field payment that becomes safe in minutes, not an exchange hold. The check scores "safe within seconds" — finality within about five seconds, or unconfirmed payments the network protects (no replace-by-fee by default, with double-spend proofs or instant locks). Fast blocks alone do not count; a chain with replace-by-fee by default fails. BTC fails here too. |
| UTXO ledger | Privacy | Section 2 ("Transactions") and Section 9 ("Combining and Splitting Value"): the paper describes value as a chain of unspent outputs that each transaction consumes and creates, with ownership proven by signatures over previous outputs. The account model keeps a running balance per address instead, which the paper does not describe. Both can work; this check scores the architecture the paper describes, which is a privacy property too: a wallet takes a new key for each payment, where an account links every payment its holder makes. |
| Privacy | Privacy: whitepaper privacy, then three protocol checks | Section 10 ("Privacy"): "the public can see that someone is sending an amount to someone else, but without information linking the transaction to anyone." The paper's own model, public keys kept anonymous and a new key pair for each transaction, is the first check: keys not tied to an identity by the chain's rules (with the UTXO ledger, the paper's new key for each payment). Network-level privacy, hidden amounts and hidden parties go further, in the base protocol: shielded transactions (Zcash), ring signatures and stealth addresses (Monero), confidential transactions. Wallet-side mixers like CoinJoin do not count. BTC passes the first two and fails the other three. |
A check not yet researched is left out and the score is scaled to the checks measured; the asset's score range (lo–hi) treats every unmeasured check as failed, then as passed, so a reader can see both ends.
Decentralisation's first criterion asks who can produce blocks, because the people who produce blocks are the people who can change the chain. Proof of work is open to anyone with hardware, and no fixed group can shut a miner out or outvote the rest. Permissionless proof of stake is open to anyone holding the coin, weighted by how much they hold. Both rank high. A capped validator set has a fixed number of seats; delegated or elected producers — 21 on EOS-style chains, a few dozen on BNB Chain — put block production in a small group; an allow list, federation, council or company decides who may produce blocks at all. These rank low. The ladder puts proof of work first and tokens issued by a founder or company last, with a point off for a permissioned list; until method version 28 an "anyone can mint" check asked the same question a second time.
The fewer the hands, the higher the risk. A small group can agree to change the consensus rules, reverse or freeze transactions, refuse a block producer, censor an address or halt the chain, and it can be pressured, bought or captured as a group. A chain that anyone can join has no such group to lean on, which makes it more robust: its rules change only when a broad and open majority goes along. That is why an asset whose chain or supply is in the hands of a small group grades lower, whatever its technology. The security of proof of stake itself — stake at risk, slashing, finality — is real and is not marked down; what is scored is how concentrated the power to change the chain is.
This ranking is the site's position on who should control a chain. It is not a claim that proof of work is more secure than proof of stake in every respect: the two face different attacks, and the security of each depends on how hashpower or stake is distributed, what it costs to acquire, and how the protocol responds to an attack.
For a mined chain, two of the nine security points are what the chain pays its own miners a day, on a scale from $1,000 (0) to $10 million (10) — the standard "security budget": what an attacker has to outspend to outpace the honest miners. It is a proxy, not an exact cost of attack, since rental capacity, hardware markets and energy prices move the real figure. It also moves with the coin's price: a higher price raises the budget in dollars, while the cost of renting the same hashrate rises only in part. It is used because it is the one input that can be measured the same way for every mined chain, whatever its algorithm.
What counts is the share of supply founders hold without having mined or validated it like everyone else, and whether they sell it. How they sell it does not matter: a presale, a private round, a public sale or sales on the market all put the founders' share into other hands for money. Developers who mined or staked on the same terms as everyone else count in neither criterion.
Premine asks who received coins without mining them; fundraising asks whether those holders sell. A foundation that keeps a premine and never sells any scores low on the first and 10 on the second. Once insiders sell, fundraising counts everything they received, not only what they have sold so far: Ripple and its founders received all of XRP and sell it, so the whole supply was and is theirs to sell, and XRP scores 0 on both. The same coins count twice on purpose, because two risks are present at once: supply concentrated in a few hands, and hands that turn it into money. The share actually sold is researched and shown on each coin's page beside the score.
The criteria fall into groups, and each asset page shows the average of each: consensus (the consensus ladders); the Cryptocurrency Standards; issuance (premine, fundraising, developer share); network security; decentralisation (holders); liquidity; and the record. An asset can score well on one and badly on another — strong security with a thin market, or an open launch with concentrated holders — and the group averages show where the letter comes from.
The grade is a measurement against a stated standard, with every input published. It is not a probability of failure, a credit rating or a forecast, and it has not been backtested: no claim is made that it predicts hacks, collapses or attacks better than simpler measures. Grades are archived every day from 24 October 2026 at /api/grades-history/, so the test can be run on the record as it grows, by anyone, against documented failures.
Grades are recomputed every six hours. The ladders and liquidity follow live data, so a score can move by a point between builds. Each grade states the build it comes from and how many criteria it rests on; an older copy of a page, in a search engine's cache for instance, may still show the previous score. The full set, criterion by criterion, is published as JSON at /api/v1/grades.json, so any grade can be recomputed from the published rules; the grade audit lays out eight major coins criterion by criterion.
What the people behind a chain have said and done, from documented facts: a launch as a joke with founders saying it has no purpose, claims by its founders or lead company that contradict its own design, and halts, freezes or reversals by a central party or coordinated validators in the last five years. Each costs 1 point and is sourced on the asset page. Court outcomes are not scored, and a chain split is not a deduction.
The standards any cryptocurrency can be held to, in six sections of the grade, opening with the qualities of sound money that Bitcoin was built to have: sound money qualities, privacy, scalability, security, programmability and decentralisation. They start from properties drawn from the Satoshi 2008 Bitcoin whitepaper, each scored where it belongs: a limited supply (Section 6, a predetermined number of coins) in sound money's scarcity, no authority that can halt the chain or freeze funds (Sections 1 and 2, proof instead of trust) and no central team (the conclusion's nodes that need little coordination) as criteria of decentralisation, a UTXO ledger (Sections 2 and 9) and keys not tied to an identity (Section 10) in privacy, anyone can mint (Sections 4 and 5) in decentralisation's who-can-mint ladder, no pre-issuance and no developer tax (Section 6, the block reward as the only way coins enter circulation) in premine and developer share, fees under one cent (the introduction's "small casual transactions") and payments safe to accept within seconds in scalability, and hidden amounts and parties in privacy. Until 27 October 2026 the criterion was called the Bitcoin Whitepaper standard; the checks are standards for any cryptocurrency that the paper states first, explained one by one in the journal post on the Cryptocurrency Standards. The standard is the DOCUMENT, not any one chain's live behaviour today — so BTC itself can fail parts of it, and does: its fees are above a cent, replace-by-fee makes unconfirmed payments unsafe to accept, and it hides neither amounts nor parties.
The two consensus ladders rank families by market value, so they follow the market: proof of work leads them because Bitcoin is the largest asset and mined chains hold the most value. If proof of stake led the market, it would lead the ladders. Decentralisation's who-can-mint ladder asks who can produce blocks, a measure of permission; security scores what an attack must outspend, which only mined chains are measured on. Both are the site's position and are stated as such.
Because the standard is the Bitcoin whitepaper, and BTC meets most of what the paper describes: its own chain and consensus, open mining, a fixed supply, no premine, no sale, no developer share, the largest security budget and the deepest market. It loses points where the live chain falls short of its own founding paper: fees above a cent, unconfirmed payments made unsafe by replace-by-fee, and no privacy in the protocol.
Turnover is daily trading volume as a share of market cap, the median over 30 days. A market cap is the last price multiplied by the whole supply; turnover shows how much of that supply the price is actually set by. Where little trades, each trade moves the price further, so a market cap resting on a trickle of volume is easier to move: over the six months to October 2026, a day's trading of 1% of market cap moved the price about 12% at under 0.2% turnover and about 1% at 2 to 5%. Order books deepen with it: in a snapshot of five exchanges in October 2026, the depth within 2% of the price rose about thirtyfold, as a share of market cap, from under 0.2% turnover to 3 to 5%, and stopped rising above 5%. For coins over $10 million the book reached its full depth by about 4% and gained little beyond 6%, where a day of trading grew from 10 to 17 times the book to 36 to 100 times. Books stopped deepening above about 5%, so 5% a day is the top of the ladder, and the score falls away on both sides. Turnover is not a forecast of calm. The same test found that coins with higher turnover were more volatile, not less, at every size, because large price moves draw trading. Above 100% a day the volume and the market cap cannot both be right, and turnover is left out; a stablecoin, whose price does not move with its trading, has no ceiling. Volume stands in for order-book depth until depth is collected for every asset. The test is published in the journal: /journal/does-more-trading-make-a-coin-less-volatile.
Premine is what founders, a company, a foundation or investors received at launch without mining it. Developer share is what they receive after launch without mining it: a share of every block reward, as with Zcash's development funds, or a one-off later mint, as with the Sia Foundation's 2021 grant. Developers who mine like everyone else count in neither. Both are scored on the share of today's supply, because the more of the supply insiders hold, the more they can sell into the market at once.
That some criteria are not yet measured and are left out of the average. Holder concentration is still being collected, so every grade is provisional for now, and some checks in the two checklists are still being researched. Each grade gives its range: the score if every unmeasured criterion and check scored 0, and if every one scored full marks.
It does not claim to. The grade measures how an asset is built and traded against a stated standard; it is not a probability of failure and has not been backtested against failures. Every build since 24 October 2026 is archived at /api/grades-history/ so that the test can be run on the record, by anyone.
Because the question is the same for every asset: how much of it depends on a group that can mint, freeze, sell or change the rules. A token scores 0 on chain criteria it cannot have, on purpose: its security is borrowed from its host chain and its issuer. Tokens are ranked separately from coins, so a token is never listed against a coin.
Its launch — how much went to insiders and whether they sold — has not been researched yet, or fewer than six of its nine network-security points have been measured, or, for a mined chain, what it pays its miners has not been measured. A grade is withheld until it is, so a coin is never rated higher for what is not yet known.
/api/v1/grades.json publishes every criterion for every asset. Average the criteria an asset has, multiply by ten, and read the band: A++ from 90, A+ from 82, A from 74, B from 64, C from 52, D from 40, E from 25, F below. The grade audit page lays out eight major coins this way.
Criteria added or restructured, by method version. Every grade is recomputed on the current version; scale changes between these are not listed.