What Is a Meme Coin Bundle? How to Check Bundle % and Linked Wallets on GMGN

In meme coin trading, a bundle usually means a group of buys placed by several wallets in a coordinated way in the first moments of a launch. Underneath, it may be executed through a transaction-packing mechanism such as a Jito bundle, or simply flagged by analytics tools based on same-block buying, shared funding sources and other on-chain patterns — which is why GMGN’s Bundle %, a Jito bundle and Bubblemaps’ bundle label are not the same metric. Check what a number measures before you decide whether it is high. What traders actually want to know: how much of the early buying was coordinated, which wallets ended up holding that supply, and whether those wallets still hold it or are linked to each other.

On GMGN, you can work through it in three steps:

1. Check the bundle percentage (Bundle %) in Trenches — it measures how much of the token’s trading came from bundled buys, so a high reading means the launch phase saw significant bundled buying;
2. Open the token detail page and filter Trades by Bundler to see whether those wallets are still holding, or already exiting together;
3. Cross-check the DEV, Sniper, Insider and Top 10 labels, then use the holder bubble map supported on that chain (InsightX, Bubblemaps, Faster100X) to test whether the large holders are linked: do they form a single cluster, can they be traced back to the deployer or a shared funding wallet, and how much do they hold combined.

What is a bundle in crypto, and why do new launches get bundled?

A bundle packs “create the token + the first buys from several wallets” into one group of transactions that executes in a set order, in the same block, with no outside trade able to slip in between. Launches get bundled because the earliest fill has the lowest cost basis, and sniper bots are watching every new mint — as long as the bundle lands on-chain, that supply ends up in a set of addresses that look unrelated to each other, at the lowest price available, and outside buyers can only enter after it has executed.

Take Jito bundles on Solana: a bundle holds up to 5 transactions, executed atomically and in the submitted order within the same slot — all of them succeed or none do (Source: Jito); while it executes, outside transactions cannot be inserted into its internal sequence (Source: Eclipse Labs on Jito-Solana). Those guarantees apply inside the bundle: once it lands, nothing external reorders it. And because “create the token + first buys” travel together, the first tranche of supply is already filled the moment the token exists, leaving sniper bots no gap to front-run.

The payoff comes from how these launches are priced: whether it is a launchpad bonding curve such as Pump.fun’s, or AMM pricing once a pool is live, the price rises as more supply is bought, so buying earlier means buying cheaper. In the MELT dataset (published in 2026, covering December 2024 to March 2025), 98.7% of token creation transactions also contained a buy from the developer — “deploy and immediately buy” is close to the default (Source: MELT — A Behavioral Trace Dataset for High-Risk Memecoin Launch Detection). That figure describes how common it is for a dev to take the first fill, not multi-wallet bundling itself; but it shows how strong the incentive to be first is, and multi-wallet bundling is the next step along the same incentive.

What traders see on the page is the result of that bundled launch. The holder list shows a dozen or more separate addresses, entering at similar times with similar sizes, looking like a crowd of independent buyers. In reality their cost basis is close to zero, and once the price runs, they can sell into everyone who arrives later. Across 41,470 launches on Pump.fun, MELT grouped related accounts using a definition broader than bundling alone (it also counts co-purchases and shared funding sources) and found that 36.5% of supply on average was held by coordinated accounts that appear, in raw on-chain data, to be unrelated addresses — note this is not GMGN’s Bundle %; see the FAQ for how the definitions break down (Source: MELT).

So checking bundles really comes down to two questions:

• Are those separate holders acting together? Was the supply bought by independent traders placing their own orders, or packed into bundled buys by a single party;
• How much of it is left, and who holds it now? That is what decides whether it can still hit the chart.

Bundling is only an execution mechanism, and legitimate arbitrage and batch operations use it too (Jito lists atomic arbitrage and similar MEV strategies among its typical use cases). The numbers above come from Pump.fun launches on Solana, and other chains and launchpads will differ; what generalizes is the mechanism — bundled buying hides real supply concentration, and concentration is what matters for trading.

Do multiple wallets buying at once mean it is bundled? Bundler vs. sniper

Not necessarily. A burst of wallets buying at launch can be bundled buying, or simply snipers racing each other — they look almost identical on the page, but the two labels measure different things: a sniper is defined by how early a wallet bought; in GMGN’s terms, a bundler is defined by whether the buys were packed together, and buying early is not the same as being flagged as bundled. (Some analytics tools use a wider definition — see the bubble map section below.)

The distinction matters because the evidence is not equally strong. Buying early only proves speed, and anyone can be fast. Multiple addresses appearing inside the same bundled buys points to coordination instead. When MELT links accounts, it treats “multiple addresses buying inside one transaction” as a single controlling entity, on the grounds that signing such a transaction requires the private keys of every account involved (Source: MELT). Open one of these multi-account buys in a block explorer and the Signer field lists every signing address.

On GMGN, Sniper marks a wallet that buys in earlier blocks after the pool is created, while Bundled Tx Wallet (Bundler) marks bundled trades — defined by GMGN as a single account combining multiple wallets’ transactions into one transaction bundle, processed in the same block (Source: GMGN — Featured Icon Definition). The two labels can overlap: the same address can be both a sniper and a bundler.

Once you know which one you are looking at, the next move is the same either way: go back to those wallets and check what they hold now and what they have done since (see the Bundler section below).

How to read GMGN’s Bundle %: what does a high bundle percentage mean?

GMGN’s bundle percentage (Bundle %) is a screening metric: it answers “does this token show a bundling signal,” not “is this token safe.” Whatever you decide after seeing the number depends on the individual bundler wallets, not on the percentage alone.

When you scan new launches in Trenches, the bundle percentage sits on the same card as the sniper ratio, insider ratio, top holder ratio and whether the dev has sold. These metrics do different jobs: the bundle and sniper ratios describe buying behavior at launch, the top holder ratio and dev sell status describe the current supply structure, and the insider ratio flags supply that may have moved to other addresses by transfer. Read together, they separate “was there concentrated buying early on” from “is that supply still concentrated now.” The other metrics come up again in the linked wallets section below; this section stays on bundling.

A bundling signal carries two layers of information:

Bundle %: how large the signal is. GMGN’s public bundler metric, bundler_trader_amount_rate, measures trading volume coming from bot-bundled buys (Source: GMGN gmgn-skills repository) — in other words, the percentage reflects how much of the trading was bundled, not how much supply those wallets hold right now. For how to read the number, and what value is worth referencing, see the section on safe bundle percentages below.

Bundler wallets: who is behind the signal. Which addresses were involved and what they are doing now takes you into the individual wallets — the next section.

How to find the bundler wallets behind a high Bundle %

Filter Trades by Bundler on the token detail page to see the individual addresses, then check what they hold now, what they have traded since, and how they relate to other flagged wallets.

Where to look:

1. Open the token detail page and switch to the Trades tab below the chart;
2. In the filter row, select Bundler — the label also shows how many bundled trades have been detected, for example “Bundler 21”;
3. The filtered list shows the buys and sells tied to those bundled trades, with the Trader column holding the wallet addresses: click an address to open its page and review current balance, holdings and trade history; click the explorer icon next to a trade to open it in a block explorer and inspect the signing addresses and token transfers.

Addresses in the bundler list also carry their other labels, such as top holder and fresh wallet, so overlaps with other categories are visible at a glance. When you need to look at holder relationships, the bubble map entry point sits to the right on the same row.

Once you have the wallets, work through this:

How to tell whether the top holders of a meme coin are linked wallets

To judge whether large holders are connected, read the wallet labels, current holdings and the bubble map together: first identify the bundler, DEV, sniper and insider addresses, then check whether any funds or tokens have moved between them.

This is worth checking whenever the dev holds very little. GMGN’s Suspected Insider label covers exactly this kind of relationship — multiple position holders that share the same creation time, funding source and transfer time (Source: GMGN — Featured Icon Definition). The dev page may show a small balance while several early wallets still sit in the top holders; dev holdings only describe the identified deployer address, so whether the other large holders are related is a separate question.

Note that top 10 concentration and wallet linkage are different things: the first answers “is supply concentrated,” the second answers “is there an observable connection between these holders.”

The bubble map is where that gets checked. GMGN provides bubble map access on the token detail page — currently InsightX, Bubblemaps and Faster100X — and which data source is available differs by chain, so go by what the page shows.

What do the bubbles and lines on a bubble map mean?

Definitions vary by provider; on Bubblemaps, each bubble is a main holder, bubble size reflects position size, and a line means the two addresses have transferred tokens on-chain, with the default view covering the main holders and connected addresses grouping into clusters (Source: Bubblemaps — How does it work?).

So when the holder list looks like a set of unrelated whales but those wallets form an obvious cluster on the map, keep going:

• which addresses have sent tokens to each other;
• whether the transfers happened before or after launch;
• the size and direction of those transfers;
• whether they share an upstream or intermediate wallet.

A cluster proves there is a traceable on-chain relationship; proving one party controls all of them takes funding history and specific transactions.

Do bundled wallets always cluster together on a bubble map?

No. Bundling and bubble map lines are two different detection methods: bundling looks at when the wallets bought and whether their buys were packed together, while a line on the map means the wallets have transferred tokens between themselves. Bundled wallets each buy from the pool and often never transact with one another, so by default there may be no lines at all.

Bubblemaps gives examples in both directions when explaining bundles versus clusters. One direction — flagged as a bundle, unconnected on the map — is exactly the case in this section’s title: within the first two blocks after a new meme coin launched, 18 wallets each bought 0.5%–2% of supply with no funding relationship between them, flagged as a bundle purely because they bought at almost the same moment. The other direction — connected on the map, not a bundle — also happens: wallets that bought at different times are never flagged as bundled, yet get grouped into one cluster because their initial SOL came from the same address (Source: Bubblemaps — Bundle vs. Cluster).

Note the difference in definitions: Bubblemaps only requires several wallets to buy within a very short window to flag a bundle, with no requirement that the trades were actually packed together or funded from one source — which is not how GMGN identifies bundled trades. The same set of wallets can be flagged by one tool and not the other.

The practical takeaway: no lines on the map does not mean the bundled wallets are unrelated. Those wallets normally need someone to send them SOL before they can buy, and the shared funding address is often the easier trail to follow. The default map may not draw that layer at all: if two large holders were funded by the same upstream wallet and that wallet is not itself a top holder, they appear on the map as two isolated bubbles. On Bubblemaps, the Magic Nodes feature exists for this case, pulling in addresses that are not top holders but connect several of them (Source: Bubblemaps — Magic Nodes). Intermediate nodes still need to be read with the address identity in mind — two whales withdrawing from the same exchange does not make them the same party.

What is a safe bundle percentage? Does a high bundle % always mean a rug?

There is no single “safe Bundle %” that applies to every meme coin. A high bundle percentage means the launch trades and supply structure deserve a closer look, but on its own it does not prove a token will rug. The percentage describes one class of bundling signal; the actual risk still depends on what those wallets hold now, how concentrated supply is, and how the wallets relate to each other.

One figure is worth knowing about: GMGN’s AI agent tooling uses 0.3 for the bundler ratio in its screening examples. It is not a safety certification — it narrows a batch of new launches down to the ones worth checking further, and does not tell you whether one particular token is safe (Source: gmgn-skills).

A ratio that looks “not that high” still needs the same case-by-case read: how much those wallets still hold, whether they are linked to each other, and what the other supply signals — DEV, insiders, snipers, the top 10 — show. The two sections above cover where to check each.

One more definitional point: Bundle % reflects how much of the trading was bundled, not how much supply those wallets currently hold; different tools also measure different objects, so the numbers are not directly comparable across platforms. GMGN’s bundle percentage is on-chain data meant to support trading decisions, not a safety rating.

Avoiding single-threshold conclusions is also how the research side works — MELT uses 122 behavioral features to classify high-risk launches, rather than any one bundle cutoff (Source: MELT).

The same study offers one result that speaks directly to bundling: after merging bundle-related accounts and recalculating top 10 holdings, the median top 10 share of high-risk tokens rose by 24 percentage points, against 6 points for low-risk tokens — what a bundling signal really points to is concentrated supply split across many addresses (Source: MELT).

So whether the question is “is 15% bundled safe” or “will a high bundle % rug,” the fuller test is the same: which wallets that percentage covers, how much they still hold, and how they relate to each other. Bundling gives you a clue about the structure of the early trades; what those wallets hold, trade and connect to now is what tells you how to read that clue.

Reading meme coin bundles is not about finding a percentage that decides a trade for you. It is about following that number down to the wallets: which addresses took part in the bundled trades, how much they still hold, and whether there is a verifiable on-chain link between the main holders.

GMGN’s Bundle % and Bundler list locate the bundling signal; the DEV, Sniper, Insider and Top 10 labels fill in the current supply picture; and when holder relationships still need confirming, the bubble maps supported on that chain — InsightX, Bubblemaps, Faster100X — take it from there.

Note: updated August 2026. GMGN’s metrics, bubble map data sources and per-chain support may change as the product is updated — go by what the current page shows.

Risk disclosure: meme coin trading carries extreme risk and can result in the total loss of your capital. This article is for informational purposes only and does not constitute investment or financial advice. Do your own research and make your own decisions before trading. GMGN makes no promise, guarantee or projection of any return. Any user gains or profitable cases mentioned are individual and highly exceptional, do not represent typical results, and are not a promise of future returns.