Fraud Detection Mechanism

Detection Model

The platform detects fraud using behavioral deviation, not static rules.

Signals include:

• Transaction velocity spikes

• Sudden value changes

• New wallet interactions

• Abnormal graph expansion

• Wallet age anomalies

Decision Strategy

Layer	Purpose
Stream Layer	Detects abnormal behavior
ML Layer	Quantifies fraud probability
Policy Layer	Determines response

This separation is intentional and mirrors production fraud systems.