Skip to content
TheChatArchiveThe Chat Archive
Grok 4.1 Fast· xaiHugging Face Frontier Reasoning Traces

Create a blockchain analytics playbook for AML: clustering, on‑chain heuristics,…

  1. User

    Create a blockchain analytics playbook for AML: clustering, on‑chain heuristics, sanctions screening, and case workflows.

  2. Assistant

    Blockchain Analytics Playbook for AML Compliance

    Introduction

    This playbook provides a structured framework for leveraging blockchain analytics in Anti-Money Laundering (AML) investigations. It focuses on four core pillars: clustering, on-chain heuristics, sanctions screening, and case workflows. Designed for compliance teams, investigators, and analysts, it emphasizes scalable, repeatable processes to detect, investigate, and mitigate illicit activity on public blockchains (e.g., Bitcoin, Ethereum).

    Key Principles:

    • Data Sources: Use tools like Chainalysis, Elliptic, TRM Labs, or open-source (e.g., Blockstream, Etherscan APIs).
    • Risk Scoring: Assign scores (Low/Medium/High) to entities based on exposure.
    • Documentation: Log all steps with timestamps, tool outputs, and rationale.
    • Regulatory Alignment: Comply with FATF Travel Rule, FinCEN SAR requirements, and local laws (e.g., EU AMLD5/6).

    Assumptions: Basic familiarity with blockchain explorers and analytics platforms.


    1. Clustering

    Clustering identifies addresses controlled by the same entity, enabling entity-level risk assessment. Goal: De-anonymize pseudonymous addresses.

    Core Techniques

    TechniqueDescriptionTools/IndicatorsAML Application
    Common Inputs/OutputsTrack funds entering/exiting multiple addresses via the same exchange deposit/withdrawal.Exchange-tagged clusters (e.g., Binance hot wallets).Link user wallets to CEX/KEX.
    Change Address DetectionUTXO-based: Identify change outputs from multi-input txs. EVM: Track low-value returns.Wallet fingerprinting (e.g., 0.001 BTC change).Cluster HD wallets.
    Co-Spend ClusteringAddresses spending to the same output in one tx.Multi-input tx analysis.Identify wallet families.
    Peeling ChainsSequential low-value sends (e.g., 0.1 BTC → 0.099 BTC).Pattern matching scripts.Detect tumbler evasion.
    Exchange AttributionKnown deposit/withdrawal patterns from CEX APIs.Crystal Blockchain, Nansen labels.Risk propagation from hot/cold wallets.

    Workflow

    1. Input: Seed address (e.g., from customer onboarding or alert).
    2. Scan: Run clustering query (radius: 1-3 hops).
    3. Validate: Cross-check with off-chain data (e.g., KYC).
    4. Output: Entity graph with cluster ID, balance, tx volume.
    5. Risk Flag: >10 addresses or >$1M volume = Medium risk.

    Example (Bitcoin): Address 1A1zP1eP5QGefi2DMPTfTL5SLmv7DivfNa (Satoshi) clusters with early mined coins via co-spend.

    Pitfalls: False positives from shared services (e.g., Lightning Network). Mitigate with confidence scores (>80%).


    2. On-Chain Heuristics

    Heuristics are rule-based filters on transaction graphs to flag suspicious patterns. Combine with clustering for entity-level alerts.

    High-Impact Heuristics

    HeuristicPatternThresholdRisk LevelRationale
    High-Velocity>10 txs/hour, total >$100K.Velocity >5x avg.HighStructuring/smurfing.
    Round-TrippingFunds sent to self-clustered address within 24h.>50% value return.MediumLayering illusion.
    Mixing/TumblingInteraction with known mixers (e.g., ChipMixer, Tornado Cash).Any tx to mixer contract.HighProven illicit tool.
    Peel ChainChain of 5+ txs reducing value by 1-5% each.>3 hops.HighEvasion.
    Gambling WashRapid in/out to gambling dApps (e.g., Stake.com).>$50K cycle <1h.MediumLegit cover for laundering.
    Rug PullCreator drains LP pool post-hype.>90% token dump.HighDeFi scam.
    Bridge HoppingCross-chain via 2+ bridges in 48h.Multi-chain volume >$1M.MediumJurisdiction evasion.
    Dusting AttackReceipt of micro-tx (<$10) from unknown.Any + subsequent peel.LowPhishing precursor.

    Implementation

    1. Setup: Define rules in analytics platform (e.g., Chainalysis KYT ruleset).
    2. Scan: Daily/ real-time on monitored entities (e.g., customer wallets).
    3. Tune: Backtest on historical illicit cases (e.g., Lazarus Group hacks).
    4. Alert: Auto-generate if score >70/100.

    Example (Ethereum): 0x... sends 100 ETH to Tornado Cash → peels to 3 addresses → round-trips to exchange = High-risk mixer use.

    Advanced: Machine learning overlays (e.g., graph neural nets) for dynamic thresholds.


    3. Sanctions Screening

    Screen for exposure to sanctioned entities (e.g., OFAC SDN, UN lists) at wallet and tx level.

    Process

    1. Data Integration:

      • Lists: OFAC, EU, UK HMT, Interpol crypto addresses.
      • Tools: Chainalysis Reactor (pre-clustered sanctions), Scorechain.
      • Coverage: 100% of blockchains; update daily.
    2. Screening Levels:

      LevelScopeFrequencyAction
      WalletDirect control of sanctioned cluster.Onboarding + daily.Block/freeze.
      1-HopDirect tx to/from sanctioned.Real-time.Alert + hold.
      2+ HopsIndirect exposure (e.g., mixer intermediary).Weekly.Investigate.
    3. Workflow:

      • Query: Match seed address against sanctions database.
      • Visualize: Tx graph highlighting red nodes.
      • False Positive Check: Verify via clustering (e.g., shared mixer ≠ control).
      • Report: Log OFAC hit with evidence screenshot.

    Example: Wallet receives 1 BTC from 1FeexV6bAHb8ybZjqQMjJrcCrHGW9sb6uF (OFAC-listed mixer). Flag as High.

    Challenges: Address reuse, privacy coins (e.g., Monero – use heuristics). Mitigate: Probabilistic matching + human review.


    4. Case Workflows

    Standardized pipelines from alert to resolution. Use ticketing (e.g., Jira, Splunk) for auditability.

    Universal Workflow Template

    Alert Trigger → Triage → Investigation → Decision → Reporting → Close
    

    Detailed Steps

    1. Triage (0-2h):

      • Prioritize: High-risk heuristics/sanctions = Immediate.
      • Assign: Analyst based on chain (BTC/ETH specialist).
    2. Investigation (2-48h):

      StepActionsOutputs
      ClusterFull entity graph.Risk score update.
      HeuristicsRun 10+ rules.Pattern report.
      SanctionsMulti-list screen.Hit/no-hit.
      Off-ChainWHOIS, OSINT (e.g., Twitter, Etherscan labels).User attribution.
      AttributionLink to fiat ramps.KYC match?
    3. Decision (24-72h):

      • Low: Monitor.
      • Medium: Enhanced due diligence (EDD).
      • High: Freeze funds, file ITR/SAR.
    4. Reporting:

      • Internal: Case summary (template below).
      • Regulatory: SAR within 30 days (FinCEN).
      Case ID: AML-2024-001
      Entity: Cluster XYZ (10 addresses, $2M vol)
      Alerts: Mixer + High-Velocity
      Evidence: [Graph PNG, Tx IDs]
      Disposition: SAR Filed
      
    5. Close & Learn:

      • Feedback loop: Tune heuristics.
      • Metrics: False positive rate <10%, resolution <72h.

    Escalation: >$10M or terrorism = Senior review + LE tip-off.

    Tools Stack: Analytics platform + visualization (Cytoscape) + case mgmt (CaseWare).


    Best Practices & Metrics

    • Automation: 80% triage via API rules.
    • Training: Quarterly simulations (e.g., mock Ronin hack).
    • KPIs: Alert volume, TP/FP ratio, SAR conversion (target: 20%).
    • Updates: Review playbook quarterly; adapt to new threats (e.g., AI mixers).

    This playbook is a living document. Customize per jurisdiction and integrate with your compliance stack for maximum efficacy. For demos or tool integrations, contact Trailblazer Labs support.

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.