“Bundled” is three different claims wearing one word
When a checker says a token was 30% bundled, it can mean any of three things, and they are not equally strong.
Several wallets bought within the same block, or the same few seconds. This is what most tools report. It is real, but a launch block is a crowded place — snipers, bots and ordinary buyers all land in it, and none of them know each other.
Several wallets received their tokens in one transaction, under one signature. Somebody paid for all of them at once. There is no innocent reading in which two strangers share a signature.
The wallets were topped up with SOL from a common source before they bought. Strong when the source is a personal wallet, weak when it is an exchange — thousands of unrelated people withdraw from the same hot wallet every day.
The first is circumstantial. The second is direct. Most of the argument about bundle percentages comes from tools reporting the first and readers hearing the second.
How often is it actually the same transaction?
We record Solana token creations as they happen. For 2,023 of them we then read the creation block and the seconds after it, listed every wallet that bought, and kept the transaction that gave each one its tokens. Then we asked a simple question: how many of those receipts carry the same signature?
Of 1,920 launch windows with at least two tracked buyers, 314 — 16.4% — contain two or more wallets that received their supply in a single transaction.
Where such a group exists it takes a median 12.8% of supply, 24.6% at the 75th percentile and 53.7% at the 90th, with a maximum of 79.5%. In 87 windows, one transaction bought a fifth of the token or more. The groups are small and tight: 208 pairs, 69 triples, 37 quads.
Roughly one launch in six, then, contains a group that cannot be explained by coincidence. Some notes on what that number is and is not:
- The denominator is windows with two or more tracked buyers, because a lone buyer cannot share a transaction with anybody. Counting single-buyer launches in the base would quietly deflate the figure.
- We follow the 30 largest buyers in each window, so this is a floor. Sharing among smaller buyers is invisible to it, and the real rate can only be higher.
- Exactly one window had every tracked wallet on a single signature — so this is not an artefact of the creation transaction itself, which would have produced that shape everywhere.
- The percentages exclude the liquidity pool and the token's creator.
The part nobody mentions: by the time you check, it has usually gone
“How much is still held?” is the natural follow-up, and the answer is bleak for anyone hoping to use it as a warning.
| What the launch bundle still holds when first measured | Share of 1,992 launches |
|---|---|
| Under 1% — already sold out | 79.2% |
| 1–5% | 10.7% |
| 5–20% | 8.8% |
| 20% or more — still loaded | 1.4% |
It is sharpest where it matters most. Of 439 launches where the bundle took at least half the supply, 342 of them — 77.9% — were already under 1% by the time anyone measured.
The lag is not in the measurement: 92.5% of those holdings readings happen within an hour of the launch window itself being read. The lag is that a token gets measured when somebody first asks about it, and that is long after it launched. A bundle check run at the moment you hear about a token is usually reading history.
Does a still-loaded bundle mean the token is about to dump? We tested it.
This is the obvious product everyone wants: if those wallets are still holding, warn me. We measured it properly — every price taken strictly after the holdings reading, so nothing that already happened could leak into the result.
It does not predict an ordinary fall.
Over the following 24 hours, tokens whose bundle still held 5% or more fell by half 28.4% of the time, against 24.8% for bundles already under 1%. That is a 1.14x lift, z = 0.66, p = 0.51, on n = 656. Something appears only at catastrophic falls — at a 70% drop it is 23.3% against 14.0% — but nine thresholds were tested, so that single result does not survive correction. We publish it as an untested lead, not a signal.
Liquidity carries no warning either. Over the same windows the median liquidity drawdown was 10.7% against a median price drawdown of 14.9%: the pool drains more slowly than the price falls, so watching it tells you what already happened.
We are publishing a negative result about our own idea because the alternative is selling it. If a tool tells you a loaded bundle predicts a dump, ask what its sample was.
How to check a launch yourself
None of this requires our tool. It requires the creation transaction, which is public.
- Find the mint's first transactions on a block explorer — the create, then the buys in the same slot and the seconds after it.
- List who bought and how much, ignoring the liquidity pool and the creator. That gives you the same-block number every checker reports.
- Compare the signatures. This is the step almost nobody takes. If two buyers appear in the same transaction, one signer paid for both.
- Look at those wallets now. Still holding, or long gone? Both are facts about the past; neither predicts the next hour.
The catch is that this only works if somebody was watching when the token was created. A creation block cannot be reconstructed from a token's present-day holders, which is why a scan run days later can describe a launch it never saw. Where we did not witness a creation, we say so rather than guessing.
What a bundle check cannot tell you
- Whether the wallets are one person. A shared transaction proves shared funding, not shared identity.
- What happens next. See the test above.
- Whether the contract is safe. Bundling is about who bought. A transfer fee, a live freeze authority or a Token-2022 trap lives in the mint and is a separate question entirely.
- That an unbundled launch is clean. Supply can be gathered over hours from separate wallets and no same-block check will see it.
What a launch-block reading does give you is the one thing a present-day holder map never can: what actually happened at the start, with the transaction that proves it.