Bubblemaps

What it is

Bubblemaps turns one question — who is holding this token, and have any of them been dealing with each other — into a picture. It draws the top 250 holders of a token as circles sized by holdings, and draws a line between any two that have transferred funds to each other. Groups of connected circles are the product: the visual claim is that a handful of addresses moving supply between themselves may be one party wearing several hats.

Three features widen the net beyond direct transfers between holders. Magic Nodes pull in addresses that hold none of the token but sit between holders — a shared funding wallet, a shared deposit address — and redraw the map with those intermediaries included. Time Nodes split an exchange or bridge into one abstract node per time window, so wallets that touched the same venue within the same window get linked even though they never touched each other. Time Travel rewinds the map to any earlier date, which is how the launch allocation of a token gets compared against its current distribution.

Around the map sit the parts that make it a place rather than a picture: a Bubblemaps Score from 0 to 100 summarising how concentrated the supply is, a token list with Trending, New and Watchlist modes filtered by chain, timeframe and launchpad, community-submitted wallet labels, an embedded TradingView price chart, and a swap panel that will buy or sell the token you are looking at from a connected wallet. There is a second, community-facing product — the Intel Desk, where token holders propose and vote on which investigation gets attention — which is governance rather than analysis and is not what this card covers.

Pricing

The V2 app is free, and you can open a map without an account at all. An account (email and password, no KYC) buys saved maps, a watchlist and the API tab.

The advanced half is not sold, it is held. The vendor's documentation lists four features as available only to holders of BMT, its own token: P&L computation, cross-chain analytics, the AI model that interprets a cluster, and maps showing the top 1,000 holders instead of 250. No holding threshold is published on that page, in the app, or anywhere else we could find, and the page itself was last touched about a year before this card was written. That arrangement is deliberately not recorded as a price on this card: a token position is not a subscription, the sum it costs changes while you read this, it cannot be put on an expense line, and it is not refundable when the feature changes.

Separately, there are real dollar prices inside the app, all of them for things a token's own team buys rather than things a reader buys: submitting community labels is quoted at $299 for up to three labels plus $89 per additional label, non-refundable, with 15% off if you pay in BMT, and the Bolts and Token Profile promotion products are priced the same way.

The REST API is a different product with a different buyer, and its prices are the only conventional subscription here: Starter at 170 a month for 100,000 credits and Standard at 850 a month for 750,000, quoted in dollars by the vendor's own plans page, with an Enterprise tier by email. Endpoints cost different amounts of credit, so credits are not requests. There is no free API tier on that page.

Data & coverage

Fifteen chains in the app — Ethereum, BNB Chain, Solana, Base, Arbitrum, Polygon, Avalanche, Tron, TON, Aptos, Sonic, Monad, HyperEVM and two newer networks — and fourteen in the API's chain enumeration, which omits Aptos. No Bitcoin, on either list.

A map is not live. The documentation guarantees that any map you open was recomputed within the last six hours, and says newly launched tokens may lag beyond that. The home page says real time. Plan against six hours.

The default map is the top 250 holders. Addresses belonging to contracts, centralised exchanges and DEX pools are hidden by default, because they connect to everything and would swamp the picture; you can reveal them one at a time. Very high-activity addresses are tagged Supernodes and are deliberately crippled: their full transfer table will not open, and two connected Supernodes are drawn with no link between them at all, to keep the computation tractable. A cluster that runs through two exchange wallets is therefore a cluster you will not see.

Clusters are computed only from transfers of tokens the vendor has chosen to index — a deliberate filter against dust-attack and fake-token contracts creating fictitious links, and equally a filter that will miss a real relationship conducted in an unindexed asset. A token's own transfers are always included in its own map whether or not it is indexed.

The Bubblemaps Score runs over the same top 250 holders, excluding exchanges, DEXs and contracts, and the documentation is unusually straight about what that means: anything Bubblemaps does not detect does not affect the score. A token that has been distributed carefully enough to defeat the clustering scores as though it were well distributed.

Integrations

The maps are most often met somewhere else: the vendor lists CoinGecko, DEX Screener, Etherscan, pump.fun and Photon among the platforms that embed them. That embed is a B2B product of its own — an iframe keyed to a partnerId with your domains whitelisted, free for a token company showing its own map, priced by enquiry for an exchange or an analytics platform.

The REST API is the other one. Authentication is an X-ApiKey header, quota is monthly credits that differ per endpoint, exceeding it returns 429, and a flat 10,000 calls per minute is enforced per IP on top. Endpoints cover maps, map screenshots, top holders, token and address metadata, and an address subgraph. It is versioned v0 and the vendor says so: "this is still an early version". The changelog shows six entries in 2026, the most recent three days before this card was written.

No MCP server — the obvious hostname does not resolve and the API returns 404 rather than an authentication error on an /mcp path. No webhooks. No published source. Addresses can be imported in bulk from a CSV, and the address list has an export button whose output the documentation calls a spreadsheet without naming a format.

Limitations

The product is an attribution, and an attribution is a judgement you cannot audit. A line on the graph means one address sent something to another. That is equally consistent with one person running two wallets, an exchange paying a withdrawal, an employer paying a contractor, an airdrop being claimed, a market maker being funded and a bridge relayer doing its job. Bubblemaps draws the line correctly and says nothing about which of those it is. The cluster is where an investigation starts, not what it concludes, and a screenshot of a cluster circulated as proof of insider ownership is a screenshot of a transfer graph.

Magic Nodes and Time Nodes widen that gap on purpose. Time Nodes connect wallets whose only relationship is having used the same exchange inside the same time window. That is a real signal in a coordinated snipe and pure coincidence on a busy venue, and nothing in the picture distinguishes the two.

The P&L number is computed from a definition you should read first. Any inward transfer counts as a buy and any outward transfer as a sell, valued at the token's price at the time. Moving your own tokens to your own second wallet is scored as a sale at market; receiving an unsolicited airdrop is scored as a purchase. The vendor documents this plainly, which is more than most do, and it still means the figure cannot be read as trading performance.

Labels are crowdsourced and paid for. Anyone may submit one with evidence, the vendor reviews it before publication, and the submitter pays. That is a defensible model, and it also means the label set is shaped by who was willing to pay to have an address named, and that labels carry no visible date for when they were last confirmed.

The four features that make the tool a research instrument rather than a viewer sit behind a token position with no published threshold. For anyone who cannot or will not hold BMT — which includes most funds and every compliance department with a policy on holding governance tokens of a vendor whose output is used in diligence — the usable product is the free one.

No Bitcoin, no order-flow, no pricing data of its own. This is a holder-distribution tool. If the question is where the volume is trading, this is the wrong instrument at any price.

Nobody on this project has used the paid or token-gated half. Everything above is read from the vendor's own documentation and its live application, dated where it appears.

Alternatives

Arkham and Nansen answer the neighbouring question — not which wallets are connected, but who owns them — and both hand you a name you cannot audit. This card's advantage over them is narrow and real: a transfer graph is a claim about transactions, which is checkable on a block explorer, where a label is a claim about the world.

Against that, both of those cover every chain and every address rather than one token's holder list, and neither gates its core view behind holding the vendor's token.

If the graph is a means rather than an end, Dune and Bitquery will give you the same transfers as rows you can join, group and re-run — more work, no picture, and a result you can explain line by line.