I’d like to analyse bitcoin blockchain in order to detect addresses used for fraud, like money laundering. Since I cannot get any addresses info that was really used for fraud, I need to analyse which kind of address have tendency to be used for fraud.

Now, I’m just wondering which parameters I should use. I’ve tried to use period of transactions and average of transactions’ amount, but it seemed not to make sense. If you have any idea, could you teach that?

  • I don't understand what you mean by 'abnormal address.' Could you give us an example of an abnormal address?
    – Nick ODell
    Commented May 13, 2016 at 5:09
  • @NickODell Ok, I'll edit it.
    – Toshi
    Commented May 13, 2016 at 5:09

1 Answer 1


Fraud and money laundering are two different things. Fraud, or accepting or spending money under false pretenses, would be very difficult to detect on the blockchain because it requires outside knowledge. The way banks detect fraud is by using the context of what was purchased, and where. The blockchain doesn't contain that information, so supplemental data would be required for detection. Most fraud would appear identical to legitimate transactions on the blockchain.

Money laundering, however, can be detected...just not very well. Mixing services are what you'd have to look for, since they would be the most obvious way to obfuscate the origin of some bitcoin. These could be detected by looking for chains of transactions with large numbers of inputs and outputs that are then spent shortly thereafter by other transactions with large numbers of inputs and outputs. Mixing services generally tumble and mix coins together like this for several hours before they finally come to rest in a UTXO.

However, there are several problems to this approach. First, mixing services are not exclusively used by criminals...not by a longshot. Many people use them simply so that whomever paid them bitcoin can't later look up how they spent it. Second, there are many other types of services that will have a very similar pattern on the blockchain. Payment processors, exchanges, and web wallets all bundle inputs and outputs together in order to minimize transaction fees. They all create similar-looking chains of transactions. Trying to detect money laundering in this fashion will result in a rather hefty paradox of the false positives, meaning it will be difficult for you to derive accurate conclusions from your findings.

  • Thx! I've got it. Btw, what do you think elliptic and other blockchain analysis services do in order to detect fraud?
    – Toshi
    Commented May 16, 2016 at 2:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.