Pattern recognition

The amounts in Bitcoin transactions (represented as Satoshi) follow Benford’s Law, contrary to claims to the contrary. We found this by analyzing one million randomly selected Bitcoin transactions. This is also the pattern of randomly selected transaction chains in the Bitcoin blockchain, if the chains are not chosen too short. This distribution arises from natural business behavior (purchase and sale, exchange, payment of salaries or profits).

Bitcoin transaction amounts in Satoshi

green: Benford
red: BTC blockchain

How does the pattern change when a human intervenes? When the transaction amounts are manually selected seemingly at random? Then it looks like below, as we found out and show on the example of a transaction chain of 832 transactions starting from transaction 1d5caeb3f5271b46bc7da1d858aa5e43b49a8f31b4f33dc9154b35c89977e765.
A human being (no matter if he is a scammer or just wants to protect his privacy well) is a bad random number generator.
As a side note, the creator of this transaction chain has transferred hundreds of other small bitcoin amounts in side transactions to the exchanges Binance, Coinpayments, Poloniex, Bitfinex, Bittrex, Kraken, Xapo, and Huobi along the long way.

Bitcoin transaction amounts with human intervention

Bitcoin Benford
green: Benford
red: human made amounts for obfuscation

Exchanges also have patterns that can be used in the analysis. For example, SilkRoadMarketplace has the following special pattern, which we did not find in any other exchange.

Bitcoin transaction amounts of SilkRoadMarketPlace

green: Benford
red: SilkRoadMarketPlace

What do we learn from this? Although there is still much room for investigation, the examples show that people or groups leave blockchain fingerprints when selecting transaction amounts that can be used in fraud analysis.

This pattern recognition is one of many analytics we use when tracking Bitcoin from fraudsters to exchanges, such as when creating expert reports.