Open data
Dataand accuracy.
Every new token SolFlag sees is analyzed on the spot, and we record how it ended after 24 hours and after 7 days. Here are the numbers, misses included, and the data so you can check them yourself.
Loading the data…
Versión en español
The data could not be loaded. Try reloading the page.
Before you read the numbers
Almost every new token dies. Of all the ones we see, more than 99% die within a day, whatever we say. So “99% of the red ones died” doesn't prove much on its own: it has to be compared with that base rate, which is always the first row.
What really shows whether SolFlag is useful: the ones still alive (the misses, below, one by one), how many we mark green (almost none yet: minutes-old tokens rarely have reasons for green) and, later on, who is behind each launch, which is what a chart can't show.
Dead = at that point it is worth less than $10,000, RugCheck flags it as a rug, more than 95% of its liquidity was pulled or, off the Pump.fun curve, it has less than $1,000 of liquidity left or less than 0.2% of what it claims to be worth (it can't be sold).
By level at first sight
| Level | Tokens | Dead after 24 h | Dead after 7 days |
|---|
Loading the data…
At 7 days only closed days count.
By signal
Each row is the tokens that had that signal (red or yellow) when we saw them. Rows with fewer than 30 tokens are faded: with so few, the percentage says little.
| Signal | Tokens | Dead after 24 h | After 7 days |
|---|
Loading the data…
What sets the survivors apart
Analysis of 6/11/2026 with the 6,040 tokens from 29/10 to 4/11 (42 alive, 0.7%; almost all with only the 24 h result, so it may change). The red or yellow level barely tells them apart. These two things do:
How big it was when we saw it
| Market cap at first sight | Tokens | Alive |
|---|
| under $5,000 | 4,298 | 0.1% (3) |
| $5,000 – $20,000 | 578 | 0.9% (5) |
| $20,000 – $100,000 | 489 | 1.8% (9) |
| $100,000 – $1M | 79 | 19% (15) |
| over $1M | 41 | 22% (9) |
Almost every survivor already had a real market in its first minute: new pools of tokens that already existed, or markets outside Pump.fun. Today the traffic light treats them like a $3,000 token on the curve; we are going to separate them (maturity-based traffic light, from 16/11).
Who paid for the launch
| Launches that wallet paid for (in the registry) | Tokens | Alive |
|---|
| 1 | 2,543 | 0.7% (18) |
| 2 – 5 | 1,590 | 0.3% (5) |
| 6 – 20 | 1,086 | 0.6% (6) |
| over 20 | 355 | 0% (0) |
None of the 355 tokens paid for by wallets with more than 20 launches is still alive. It is a weaker signal, but it points at what we want to measure: who is behind. We will give it more weight with the wallet registry (from 9/11).
Who is behindthe launches.
For every new token we store which wallet paid for its creation, which one created it, who funded it and who bought in the first blocks. This is what shows up when you put it all together, since 30 October. Figures from 9/11/2026.
18,141launches stored
7,916different wallets paid for them
225wallets paid for 30% of them all
The repeat launchers
One in four wallets has paid for more than one launch, and together they account for 64%. The survivors come mostly from people launching for the first time.
| Launches paid for | Wallets | Launches | Alive at 24 h |
|---|
| All | 7,916 | 16,765 | 1.2% (184) |
| Only 1 | 5,959 | 5,959 | 2.2% (118) |
| 2 to 9 | 1,732 | 5,837 | 0.6% (34) |
| 10 to 49 | 214 | 3,781 | 0.9% (30) |
| 50 or more | 11 | 1,188 | 0.2% (2) |
“Alive” is calculated over those that already have a 24 h result. “Only 1” means once since 30/10: they may have launched before.
Known faces
SolFlag warns you when whoever is behind a launch already has a history. This is how those launches ended:
| At first sight | Launches | Alive at 24 h |
|---|
| No known face | 13,997 | 1.6% (203) |
| Known face, yellow | 1,883 | 0.6% (10) |
| Known face, red | 2,261 | 0.6% (11) |
The wallets that paid for the most launches
The first buyers
We have seen 3,136 wallets buying in the first blocks of 3,649 tokens (among those we analyse, up to 5 per token). The most frequent one was in 652 of them.
These are facts, not accusations. Paying for a launch does not prove who keeps the money: it may be a service launching for others, and a dead token is not a scam in itself. Something off? Write to us and we will fix it.
The ones still alive
Tokens that were not dead at their latest result. If we marked them red or yellow, they are our misses (or their time hasn't come yet).
| Day | Token | At first sight | Result | Worth |
|---|
Loading the data…
Day by day
| Day | Tokens | Red: dead 24 h | Yellow: dead 24 h | Status | Data |
|---|
Loading the data…
Download the data
One file per launch day (UTC date) and an index with the summary. They are JSON. Each day is published twice: when all its tokens have the 24 h result and, closed, when they have the 7-day one.
Format (version 1)
Each day file is { v, day, final, updated, tokens }, and tokens is an object keyed by token address. Short keys because the registry grows every day:
- m
- token address (mint)
- s
- symbol
- t
- when it was created (ms since 1970, UTC)
- d
- dex where we saw it
- p
- launch platform according to RugCheck
- a
- when SolFlag analyzed it (ms)
- l
- level at first sight: HIGH, MEDIUM, LOW, SPAM, NO_DATA, NO_MARKET
- v
- severity (number)
- ty
- token type
- f
- signals that did not pass: “CODE:b” red, “CODE:w” yellow, “CODE:n” not checked
- liq / mc
- liquidity and market cap at first sight (USD)
- w1 / w10
- % held by the largest wallet and the top 10 (people)
- py / fu / cr
- who paid for the launch, who funded it and who created it
- rug
- RugCheck flags it as a rug
- cara
- known-face level, if any
- nrc / rcl
- RugCheck did not answer at first sight / answered later and p, cr, fu and rug were filled in
- ev
- engine version that analyzed it (no ev: 0.4.6)
- h24 / d7
- result: { mc, liq, dead, cv, pull, dry, dx, gr } (cv still on the curve, pull liquidity pulled, dry no liquidity or almost none next to its market cap, dx main dex, gr graduated from the curve)
The index is { v, updated, days }; each day has levels and sig (by level and by signal: n, h24, dead24, d7, dead7), curve (curve tokens and how many graduated), alive (the ones alive) and final. If anything changes in an incompatible way, v goes up.
License
What SolFlag computes (the level and signals when first seen, how each token ended up, who paid for, funded and created each launch, known faces and the wallet base) is published under the CC BY-NC 4.0 license: you may use, copy and adapt it freely for non-commercial purposes, crediting “Data: SolFlag (solflag.com)”. For commercial use, write to solflag.team@gmail.com.
Market figures (liquidity and market cap: liq, mc and those in h24 and d7) and RugCheck's platform and rug flag come from DexScreener, GeckoTerminal and RugCheck: they are not ours and follow their own terms.
If you see a mistake
These are facts visible on the chain (who paid, who funded, how it ended), never labels like “scammer”. If an address is yours or you think something is wrong, write to solflag.team@gmail.com: we check it and, if it is a mistake, we fix it and say so on the blog, as we already did with the 31/10 numbers.
On 31/10 from 01 to 15 h UTC RugCheck did not answer and many tokens were left without creator or platform: that day has more “medium risk” than usual and the creator signals count for less.