The risk model is a sophisticated system designed to calculate risk scores for blockchain addresses based on their associations with various risk categories. The model uses a configurable parameter system that allows compliance officers to fine-tune risk assessments according to their specific requirements.
Ext - ECS Risk Model Parameters define the top-level sensitivity of risk detection, and the Risk Model Configuration System determines the priority for handling alarms during risk scoring.
The system is built around the RiskModelParams object, which serves as a centralized configuration container for all risk calculation parameters. The model supports:
| Parameter | Default | Description |
|---|---|---|
default_weight |
10.0 | Fallback risk weight for uncategorized tags |
hop_decay_gamma |
0.95 | Global hop decay rate (retains 95% risk per hop) |
default_volume_coef |
0.5 | Conservative fallback when volume data is missing |
direction_factors: {
"RECEIVE": 1.0, # Risk multiplier for receiving transactions
"SEND": 1.0 # Risk multiplier for sending transactions
}
Each risk category can be individually tuned with four key parameters:
CategoryRiskProfile = {
"weight": float, # Base severity weight (0-100+)
"hop_decay": Optional[float], # Per-hop decay rate (0-1)
"min_score": Optional[float], # Minimum risk floor
"default_volume_coef": Optional[float] # Volume coefficient when data missing
}