Add Modern AML Typologies to Alert Pattern Library

[matheusybsen]

I'm an AML analyst using AMLSim to generate test data for ML models. The current alert
patterns focus on traditional typologies (fan-in, fan-out, cycles) but are missing modern
money laundering schemes that we see in real-world investigations.

For example, the current patterns don't include:

• Trade-based money laundering (TBML)
• Cryptocurrency integration schemes
• Nested structures (multiple layering levels)
• Micro-structuring (many small transactions just below thresholds)

This makes it hard to test detection models against emerging threats.

Solution:

Add new money laundering pattern types to the alert pattern library:

1. Trade-Based ML: Over/under invoicing patterns with corresponding trade transactions
2. Cryptocurrency Integration: Cash → Crypto → Cash cycles
3. Advanced Structuring: Multiple accounts making coordinated small deposits
4. Nested Layering: 3+ levels of intermediary accounts before integration
5. Cross-Border Shells: Transactions through multiple jurisdictions with shell companies

These should be added as new alert pattern templates that users can configure in
alertPatterns.csv.

Alternatives

• Manual pattern creation: Could manually create these patterns, but it's time-consuming
  and error-prone
• Modify existing patterns: Current patterns are too simple to represent these complex schemes
• Use real data: Can't use real data due to privacy regulations

Additional context

Why this matters:

• FATF Red Flag Indicators include these typologies
• FinCEN advisories highlight trade-based ML and crypto risks
• Realistic test data improves model performance on real investigations

Pattern references:

• FATF Trade-Based ML Report (2020)
• FinCEN Cryptocurrency Advisory (2019)
• Wolfsberg Correspondent Banking Principles

Impact:

• Better ML models trained on realistic patterns
• More comprehensive testing coverage
• Helps researchers address current AML challenges

I can provide detailed pattern specifications and examples from regulatory guidance
if helpful!