Every number on a token page is rated against a published rule. This page explains each metric, where its data comes from, the rule that rates it, and the limits of free data.
What the ratings mean
Each rated number gets a five-step strip: very low, low, neutral, high or very high. The word describes the size of the number. The colour says whether that size is good or bad for a holder, so a very high fee multiple shows red because the token is expensive, while a very high share of fees reaching holders shows green. Hover or tab onto any rating on a token page to see the exact rule behind it.
Three kinds of yardstick are used, each chosen to suit the metric. Fixed thresholds apply to numbers that mean the same for every token, such as how much of the supply is still locked. Peer percentiles apply to valuation: a token's fee multiple is ranked within its own category, because a multiple that is cheap for an exchange can be expensive for a chain. Trends apply to activity, comparing the latest period with an earlier one, because a growing project and a shrinking one can show the same number today. Some numbers are shown but not rated, where a rule would only be opinion dressed up as fact.
The thresholds are house rules: published opinions, not evidence. They are starting points and will be tuned as ratings are reviewed against tokens with a known history. Each section of a token page also gets an overall verdict (strong, good, mixed, weak or poor) from the average of its rated metrics.
Informational only, not financial advice. Ratings follow published rules and can be wrong when the source data is wrong. Check anything important against the project's own documents.
How a token gets its type
A fixed rule, checked in order, decides the type, and the type decides which areas apply. A memecoin has no business to analyse, so only its supply, holders, market and contract are checked. A chain whose protocol side also earns fees, such as Hyperliquid or Arbitrum, stays a chain but also gets the value-accrual area, and its fees are taken from the protocol, because that is the business behind the token.
Coverage
The coverage badge shows how complete a page is: the share of the metrics that apply to the token that actually have data. Full means at least 80%, Partial at least 45%, and Market data only anything less. Memecoins are always market data only. Missing data is shown as "No data", never as a low rating, and a metric that does not apply to a token (such as holders for a chain's native coin) says why instead.
No record is not the same as none. Funding rounds, exploits, audits and GitHub links come from DefiLlama's and CoinGecko's lists, and those lists are incomplete. When a list has nothing on file for a project, the page says so ("Not in DefiLlama's records", rated "Unknown"), counts it as missing data, and links to free places to check by hand. It never shows "none" and never rates the gap. A check that actually ran and found nothing, such as GoPlus scanning a contract, is a real result and is shown as one.
Which tokens can be searched
The search only offers real tokens: the CoinGecko top 300 by market cap, plus any token ranked up to 1,500 that DefiLlama tracks as a protocol or chain. Fake copies that reuse a well-known ticker never appear. Any other token can still be opened by pasting its contract address; its page is labelled as outside the verified list so you can check it is the token you mean. Every token page shows its official contract addresses so they can be compared with your wallet or exchange.
Where the data comes from
All data is free and needs no account or key. Prices, supply and contract addresses come live from CoinGecko. Fees, revenue, value locked, unlock schedules, treasuries, funding rounds, exploits and category benchmarks come from DefiLlama. Holder lists and contract risk flags come live from GoPlus, read on the token's home chain. Development activity comes from GitHub. Attention comes from CoinGecko's trending searches and watchlist counts.
Most numbers are fetched live in your browser when you open a token. Your browser keeps them for a while so reopening a token is instant: prices for up to 15 minutes, other live data for up to 12 hours. The line under the token's name says how old they are, and Refresh fetches everything again. Nothing is shared with other visitors. Data that is too heavy for a browser, or not available to one, is prepared once a day by an automated job: the searchable token list, unlock schedules, treasuries, exploit history, funding rounds, development activity and the peer benchmarks. The job checks its results before saving them and never replaces good data with a suspicious result. The footer of the search page shows when it last ran and whether every source answered.
Limits of free data
DefiLlama's figure for revenue reaching token holders can miss value returned outside the protocol, such as buybacks run by a foundation, so a zero there deserves a check of the project's own documents. Its audit records are patchy, which is why audits are shown but never rated. Unlock schedules are DefiLlama's documented schedules; supply without a published schedule is counted as locked beyond twelve months. The ten largest wallets often include exchanges, bridges, staking contracts and treasuries, which hold tokens on behalf of many people, and on Solana the type of each wallet is not identified. Development activity only sees public repositories, so teams that build in private look quieter than they are. TVL counts public DeFi deposits only. For a DeFi protocol that is the business itself, but for a chain it can miss the main use (payments, enterprise or real-world-asset settlement): when a chain's TVL is under 1% of its market cap, its TVL figures are shown but not rated. Peer comparisons need at least eight projects in a category; smaller categories fall back to a wider group, and the page says so. Attention is a stand-in for social chatter, not Twitter data: X charges for reading posts, so the page uses CoinGecko's free trending list (checked once a day) and watchlist counts (recorded weekly). History started in October 2026, so the 30-day figures fill in over the first month.
The metrics
Grouped by area, in the order they appear on a token page. Each entry gives what the number is, why it matters, where it comes from and the rule that rates it.
Loading the metric list…