Chain analysis

Every transaction since the first block on January 3, 2009 is still public and still downloadable, so the work is retroactive: a payment made a decade ago can be unpicked by a method invented last year. Firms sell this as software to exchanges, tax authorities, and police forces. For a buyer it means the privacy of a purchase is decided by what you do afterwards, not at the moment you press buy.

How it works

Chain analysis builds a map in three layers, and only the first one is about bitcoin.

Heuristics turn raw transactions into probable ownership. The strongest is common input ownership: when a transaction spends several outputs at once, whoever built it held all of those keys. Next comes change detection, which picks out which output came back to the sender by looking for the odd remainder against a round payment, a script type that matches the inputs rather than the recipient, or an output that a later transaction spends in a familiar pattern. Address reuse needs no heuristic at all and remains the single largest leak, because one address paid twice publishes its own history.

Tagging attaches names to the resulting clusters. Analysts deposit small amounts to exchanges to learn which addresses belong to which venue, scrape donation addresses posted publicly, and absorb address lists published in indictments and seizure notices. The result is a graph where most large clusters carry a label such as a named exchange, a payment processor, or a sanctioned entity.

Off-chain joins finish the job. The identity document you gave an exchange, the withdrawal address recorded against your account, the IP address that first broadcast a transaction, and the block explorer you used to check your own balance are all outside the protocol and all more revealing than anything in it.

Nothing here requires special access. The heuristics are published, the ledger is roughly 700 gigabytes of freely available data, and a determined individual with a full node can reproduce most of what a commercial tool does, more slowly.

Why this matters when you buy bitcoin

The purchase itself is where identity and address get written down together, and everything downstream inherits that link.

An exchange holds your identity documents and the address you withdrew to. Every venue we review that serves customers in regulated markets does this, because it is the law in the great majority of the 231 countries covered here, not a policy choice. The transfer rules go further: under the EU regulation applying from December 30, 2024, providers must collect and pass on sender and recipient details for crypto transfers between them, and verify ownership for transfers above 1,000 euros involving a self-hosted wallet. In the United States the equivalent recordkeeping threshold sits at 3,000 dollars.

What you control is everything after the withdrawal. Take a fresh receive address for every payment; your wallet generates them by the thousand for exactly this reason. Keep coins bought with identity documents separate from coins acquired any other way, and do not merge them in a single transaction unless you accept that the merge is a public statement of common ownership. Use coin control so your software cannot make that decision for you. Run your own node or use a wallet with client-side block filters, so you are not handing your entire address set to someone else's server every time the app opens.

Be realistic about mixing. The tools a 2021 guide would have recommended largely no longer exist: the Wasabi coordinator run by zkSNACKs shut down on June 1, 2024, and Samourai Wallet's founders were indicted in April 2024. What remains in mainstream software is quieter, chiefly payjoin and silent payments, which improve privacy for individual payments rather than trying to break existing links.

How 94,636 bitcoin were traced to a 2016 hack

On August 2, 2016, 119,754 bitcoin were stolen from Bitfinex. The coins sat mostly untouched, which is exactly the condition chain analysis needs.

Investigators watched the cluster for years as small portions moved through a darknet market, were split across roughly 2,000 addresses, and were converted into gold, prepaid cards, and other assets. On February 8, 2022 the Department of Justice announced the arrest of Ilya Lichtenstein and Heather Morgan and the seizure of about 94,636 bitcoin, valued near 3.6 billion dollars at the time and described as the largest financial seizure in the department's history. Lichtenstein was sentenced to five years in November 2024.

The detail worth remembering is how it closed. The chain narrowed the field and produced the target, but the decisive evidence was a file in Lichtenstein's cloud storage account listing 2,000 addresses along with their private keys. Chain analysis rarely names anyone by itself. It builds a list of suspects and a timeline, and conventional investigation does the rest, which is precisely why an ordinary buyer's exposure comes from the ordinary records they leave, not from cryptography.

Chain analysis vs pseudonymity

Pseudonymity is a property bitcoin has; chain analysis is the industry that erodes it. The protocol never asks who you are, so addresses function as pen names. Chain analysis exists because pen names in a permanent public ledger are a weak disguise: they can be grouped, watched, and eventually matched to a person through a single leaked link. One is the starting condition, the other is the pressure applied to it over time.

Chain analysis vs KYC

KYC is a legal obligation placed on a company; chain analysis is a commercial service anyone can buy. KYC happens once, at the door, when an exchange collects your identity documents before letting you trade. Chain analysis is continuous and applies to the coins forever, including after they leave that exchange, and it works on people who never registered anywhere. They are strongest in combination: KYC supplies the name, chain analysis supplies everything the coins did next.