Every grade on Before It Rugs comes from a fixed set of rules applied to public data. The same rules apply to every coin. Nobody picks grades by hand, and nobody can pay to change one.
This page explains exactly how it works, including what the checks can't see.
In short: we add up warning points from on-chain evidence. More points means a worse grade. A good grade means we didn't find the signs we look for. It never means a coin is safe.
Where the data comes from
| Source | What we use it for | How often |
|---|---|---|
| DexScreener | Finding new and promoted coins, plus price, market cap, liquidity, volume and buys and sells | New coins every 30 seconds, market data every 10 seconds |
| GeckoTerminal | Finding new and trending pools | Every 30 seconds |
| RugCheck | Token report: mint and freeze authority, top holders, insider wallets, liquidity locks, creator history, risk flags | Re-checked every 10 minutes per coin while it is on the live board |
All three are free public sources. We're not affiliated with any of them, and they don't review or approve our grades.
Coins leave the live board after about 8 hours, unless they are still trading more than $50,000 an hour. After that we stop re-checking them, and their page shows the grade from the last check.
A question mark means we have not received its safety report yet. We request reports once a coin has at least $3,000 of liquidity (or $3,000 of trading in the last hour on a launch curve), and RugCheck limits how many we can request each minute, so a new coin can wait a few minutes. Smaller coins stay unchecked unless you paste their address into the search box. We don't grade on market data alone.
How a grade is calculated
Each warning sign we find adds severity points. Every point comes with a line of plain English on the coin's page, so you can see why it scored what it did.
The signals
| Signal | Points | What it means |
|---|---|---|
| RugCheck marks the token as rugged | 9 | Liquidity was pulled or the token was abandoned |
| Mint authority still active | 4 | Whoever holds it can create new tokens and sell them |
| Freeze authority still active | 4 | Whoever holds it can stop wallets from selling |
| Identical bags (3 or more top wallets) | 3 | Same-size bags in several top wallets usually means one team bought across many wallets at launch (a bundle) |
| Insider wallet cluster, 20 or more linked wallets (counted only when linked wallets are at least 3% of holders, or the coin has fewer than 300 holders) | 3 | Wallets funded from the same source, acting as one holder |
| Insider wallet cluster, 10 to 19 linked wallets (same 3% or 300-holder condition) | 2 | As above, on a smaller scale |
| One wallet holds 15% or more of supply | 3 | A single seller this size can crash the price alone |
| One wallet holds 8% to 15% of supply | 2 | A single large holder can move the price a lot |
| Top 10 wallets hold 50% or more | 2 | Supply sits with very few hands |
| Top 10 wallets hold 30% to 50% | 1 | Supply is fairly concentrated |
| Liquidity not locked | 2 | Whoever added the liquidity can withdraw it |
| Creator has launched tokens that rugged | 3 | The creator wallet has a history of failed or rugged launches |
| Creator still holds 5% or more | 1 | The creator can sell into buyers |
| Other danger flag from RugCheck | 2 each, up to 2 flags | A serious risk RugCheck flagged that isn't covered above. RugCheck warnings about liquidity, mint, freeze, ownership, holders, LP unlocks, the creator or copycats are skipped, because we score those ourselves |
| Very thin liquidity (under $5,000, coins trading in a normal DEX pool only) | 1 | Small sells move the price a lot, and you may not get out at the price you see |
| Copycat of an older coin | 2 | Another coin with the same ticker came first |
Bundles, insiders and big holders score 1 to 3.
A few details that matter:
- Holder figures leave out pools and lockers. Liquidity pool accounts and lock contracts aren't real holders, so we remove them before measuring concentration.
- Identical bags means at least 3 of the top 20 wallets each hold 1% or more, with bag sizes within 0.05 percentage points of one of those wallets' bags.
- Insider clusters use RugCheck's count of linked insider wallets. On large coins that count can run into the thousands for ordinary reasons, so it only adds points when linked wallets make up at least 3% of holders, or the coin has fewer than 300 holders.
- Older tokens. The live board covers coins launched in the last 14 days. Older tokens can still be checked on request by pasting their address into the search box. For tokens more than 14 days old we score only RugCheck rugged status, mint authority and freeze authority, plus the very thin liquidity and copycat checks that apply to every coin. Holder checks are not scored, because exchange and treasury wallets hold large, similar-sized bags for ordinary reasons.
- Thin liquidity only applies once a coin trades in a normal DEX pool. Coins still on a launchpad bonding curve (such as pump.fun) don't have pool liquidity yet, so this check waits.
- Where two bands of the same signal exist (for example, top 10 at 30% and at 50%), only the higher one counts.
From points to grade
| Points | Grade |
|---|---|
| 6 or more | F |
| 4 to 5 | E |
| 3 | D |
| 2 | C |
| 1 | B |
| 0 | A, if the coin also has 500 or more holders and is at least 30 minutes old. Otherwise B |
The extra conditions for an A exist because a coin a few minutes old with a hundred holders hasn't had time to show its hand. Zero points on a brand new coin earns a B until it's older and more widely held.
A worked example
A coin with active mint authority (4 points) and unlocked liquidity (2 points) has 6 points, so it's graded F. Its page lists both reasons in plain English, worst first.
If the creator later revokes mint authority, the next RugCheck report drops those 4 points. The coin then has 2 points and moves to C. Grades follow the evidence, so they change when the evidence changes.
Copycat detection
Copycats reuse the ticker of a coin people are already talking about, hoping you'll buy the wrong one.
We normalise every ticker by making it lower case and removing anything that isn't a letter or a number. So "$KIMCHI", "Kimchi" and "KIM-CHI" all become "kimchi". When two or more coins we're currently tracking share the same normalised ticker, the one whose trading pair was created first counts as the original. Every later one is marked as a copycat and gets 2 points.
To avoid flagging dust, we only compare coins with a market cap of at least $10,000.
Being the oldest doesn't make a coin legitimate. It only means it came first. And a copycat isn't always malicious. The flag is there so you check you've got the coin you meant.
Paid promotions
We also note when a coin pays for promotion on DexScreener, such as a boost or a paid profile, and record the price at that moment. This is shown on the site so you can see what happened afterwards. It doesn't add or remove points.
How we track outcomes
When a coin is first graded E or F, we record its price at that moment. From then on we show how far the price has moved since we flagged it. We publish these results, good and bad, so you can judge whether our flags are worth paying attention to.
The survivors leaderboard does the opposite. It lists new coins at least 24 hours old (with 3-day and 7-day cohorts) that are still standing: at least $5,000 traded in the last 24 hours, at least $5,000 of liquidity (or still on a launch curve), and not down 90% or more from the highest price we recorded. It only includes coins we first saw within 6 hours of launch and under $10 million market cap. Surviving isn't a recommendation. Plenty of coins survive a week and then collapse.
Tokenised stocks, wrapped tokens and stablecoins are not memecoins, so they are left out of the board, the survivors and the record.
What on-chain checks can't see
Our checks read what's on the blockchain and in public reports. Lots of rug pulls happen in places that data doesn't reach.
- Intentions. We can see what a creator could do, not what they plan to do.
- Hidden coordination. Insiders can buy through fresh wallets funded from exchanges, which hides the links between them.
- Slow exits. A team can sell a little at a time over days without tripping any single check.
- Off-chain promises. Fake partnerships, paid influencers, faked socials and staged "community takeovers" aren't visible in token data.
- Locked liquidity isn't the whole story. Even with liquidity locked, large holders can still dump their tokens into it.
- Timing. Market data is up to 10 seconds old and token reports up to 10 minutes old (sometimes longer when rate limits bite). A lot can happen in 10 minutes.
- Coverage. We find coins through public feeds. We don't promise to catch every new Solana token.
- Source errors. If DexScreener, GeckoTerminal or RugCheck report something wrong, our grade will be wrong too.
That's why an A is not a safety rating. For more, read can a safe-looking coin still rug? and what our grades mean.
How we handle errors
- If a source goes down, we keep showing the last data we had, with the time it was fetched, rather than guessing.
- If we have no token report, the coin shows as not yet checked. We don't grade on market data alone.
- If a rule is wrong, we fix the rule, re-grade affected coins and note the change on this page with the date.
- If you spot a mistake, please tell us. The corrections policy explains what happens next, and you can reach us through Contact.
Changes to this method
When we change the signals, points or grade bands, we update this page and the date at the top. Grades shown before a change were calculated under the rules in place at the time.
- 21 October 2026: RugCheck warnings that duplicate our own checks (liquidity, mint, freeze, ownership) are no longer counted twice. Some grades went down as a result.
Nothing on this page is financial advice. Please read the risk warning.