How the score works

Every token gets a score from 0 to 100, and higher is safer. It's built from six signals with fixed weights, and the weights are the same for every token.

The six signals

Each signal is scored by its own checks and then multiplied by its weight. A token that passes every check in a signal gets all of its points, and loses them gradually as the risk grows.

Grades

Flags that cap the score

Some problems outweigh everything else. If a token has one of these, its score can't go above the cap, however good the rest looks:

• A permanent delegate can move tokens out of any wallet

  Contract

  Max 30

• Mint authority is still active

  Contract

  Max 30

• Freeze authority is still active

  Contract

  Max 30

• Wallets that bought at launch still hold over a quarter of the supply

  Launch

  Max 35

• Bundled wallets still hold over a fifth of the supply

  Launch

  Max 35

• One holder owns over 30% of the supply

  Insiders

  Max 40

• A transfer hook could block sells

  Contract

  Max 45

• Most of the liquidity can be withdrawn at any time

  Liquidity

  Max 45

• If the top 10 holders sold, the price would fall by over 95%

  Liquidity

  Max 45

• Wallets linked by token transfers hold over half the supply

  Transfers

  Max 60

When several apply, the strictest cap counts, and the scan lists that flag first among its reasons.

When a check can't be read

Points are earned, never assumed. Sometimes a check can't run: a data source might not answer in time, or a brand-new token may not have a trading pool yet. That signal is then marked Not checked and earns no points, and nothing is scaled up to cover the gap. Anything that couldn't be read, like whether metadata can change or how much of a pool is locked, counts as a risk in the same way.

A missing check never triggers a cap. Caps only apply to facts that were actually read.

Some checks matter too much to skip: the launch, the holders and the liquidity. If any of them couldn't run, the score can't go above 74, so the grade is Caution at best. Whenever a check didn't run, or ran on partial data, the scan shows a Limited data note that names it.

What each signal looks at

Contract and tech (20 points)

Six checks on the token's contract:

• Mint authority revoked: no one can create new tokens.
• Freeze authority revoked: no one can freeze wallets so they can't sell.
• Metadata locked: the name, ticker and image can't change.
• No transfer fee: transfers aren't taxed. A fee above 5% counts as serious.
• No transfer hook: no custom code runs on every transfer.
• No permanent delegate: no other account can move tokens out of your wallet.

Each failed check takes points away. An active mint authority or a permanent delegate costs the most.

Block zero, snipers and bundles (20 points)

Unmask reads the token's first transactions and finds block zero: the first block where anyone could buy, which is the launch itself on a launchpad, or the moment liquidity was first added. It then finds the wallets that bought in that block and the four after it. What matters is how much of the supply those wallets still hold: it starts to count at 3%, is serious from 8%, and leaves no points at 10%.

A bundle is a group of launch buyers with a proven link: they were funded by the same wallet (not an exchange or other busy service, which funds thousands of unrelated wallets), or one transaction bought for all of them. Wallets that only bought in the same block are shown, but don't count, since that's also what a busy launch looks like. Bundles are scored on the share of supply they still hold, and the signal is the average of the two.

Only tokens bought from the bonding curve or a pool count as buys. The initial mint and plain transfers between wallets don't.

Insider concentration (15 points)

The share of supply held by the 10 largest holders, leaving out liquidity pools, burned tokens, exchanges and programs. A top 10 holding 10% or less gets full points. It starts to count as a risk at 25%, and from 40% there are no points left.

The dev's own wallet and wallets the dev funded are counted on their own as well. Together they start to cost points above 2% of supply, and more above 5%.

Liquidity and exit slippage (15 points)

Two checks, averaged:

• Liquidity: read from the pools themselves: each pool's vaults on-chain, valued at Jupiter's prices. Only pools you could sell into for SOL, a stablecoin or another liquid token count. Depth is the pool against the market cap (10% or more gets full depth points, 2% or less none), or its size alone for big pools ($2.5M or more is full depth). The rest of the points are for the share of liquidity that's burned, locked or on a bonding curve; liquidity that can be withdrawn, or whose lock status is unknown, earns none of them.
• Exit slippage: Unmask asks Jupiter to quote real sells and measures how far the price would drop if someone sold 1% of supply, and if the top holder sold their whole bag. A 1% sell starts to matter at a 35% drop and is serious from 55%. The top holder selling starts to matter at 50% and is serious from 80%.

The scan also shows what would happen if the top 10 holders all sold. It doesn't change the points, but a drop of over 95% caps the score (see the flags above).

Holder behaviour and sell pressure (20 points)

Unmask matches the largest holders against FOMO usernames, then reads each matched trader's complete trading history on-chain, across every wallet FOMO confirms is theirs, to see how they usually trade: how long they hold, and how often they sell half or more soon after buying.

Each trader gets a likely next move: the chance they sell half or more of this bag within the next hour, from a model trained on traders' complete histories (it's only used where it beat the badge rules on recent data it hadn't seen) or from how often traders with the same badge sold at that point. A trader without enough history to judge counts as a risk: their whole bag. Each chance is weighted by the size of their bag, and added up, that gives the token's sell pressure:

• Low: under 1.5% of supply.
• Medium: from 1.5%.
• High: from 4%.

If none of the FOMO traders Unmask can see hold a large bag, sell pressure is Low.

Suspicious transfers (10 points)

Plain transfers are a common way to spread one big position across many wallets so it looks smaller. Unmask looks for supply held by wallets that are linked to each other through token transfers, using RugCheck's insider network data. It starts to count at 2% of supply and is serious from 6%. If linked wallets hold half the supply or more, the whole score is capped at 60.

Trader labels

Each FOMO trader in a scan gets short labels that say what their own trades show. A label about habits needs at least five recent trades, and links to the trades behind it:

• Sold half or more within 10 minutes, when they did that on at least 2 in 5 of their recent trades.
• Held for a week or more, on the same 2 in 5 rule.
• Got this token by transfer, without buying it.
• No strong pattern when their trades don't point either way.

Labels describe activity. They aren't accusations and say nothing about anyone's intent.