SC
Sarah Chen
Retail Client · RC-2024-08821 · Onboarded: 2019-03-14
Vancouver, BC
Restaurant Server
$38,000/yr
6.8 years
Chequing, Savings, CC
$4,200 (last 3m)
$28,700 ▲
Retail-FoodService-BC
Composite Score
0.94
High anomaly across
3 of 4 foundational lenses
3 of 4 foundational lenses
Contextual Anomaly Score
CWAE-MMD0.97
99th percentile among Retail-FoodService-BC peers
Context-Adjusted Percentiles (behavior | profile)
Wire Volume
P96
Cash Deposits
P92
Cross-border
P88
Txn Velocity
P85
Avg Amount
P41
Interpretation: This client's wire transfer volume and cash deposits are extreme outliers given their declared income ($38K) and occupation (food service). The model expected ~$2,100/mo; observed $28,700 in 3 months.
Feature Attribution (Why This Score?)
DynamicLRP
Signed relevance — red = drives anomaly up, green = model expected normalization
wire_amount_3m
+0.34
cash_deposit_count
+0.28
cross_border_ratio
+0.18
txn_velocity_1m
+0.12
unique_counterparties
+0.07
avg_txn_amount
-0.05
account_tenure
-0.03
Conservation: ΣR = 0.91 (target: 0.94) — 96.8% conserved
Context fraction φc = 8.2%
Interpretation: 82% of anomaly driven by wire transfers + cash deposits. Negative relevance on avg_txn_amount means the model tried to normalize (individual txns are moderate), but volume is extreme. Low φc confirms the anomaly is in behavior, not context leakage.
Transaction Network Pattern
GNNSC
A1
A2
A3
A4
A5
A6
B1
B2
B3
Subgraph: 10 nodes · 14 edges · 3-hop ego
Pattern: GATHER-SCATTER (conf: 0.89)
Alert Client
Source (inflow)
Destination (outflow)
Suspicious edge
Pattern: GATHER-SCATTER — multiple sources funneling into client, then dispersing to fewer destinations. Consistent with layering or structuring behavior.
AML Typology Mapping
COMPOSITETypology match scores (multi-lens consensus)
Layering
0.91
High Match
Structuring
0.78
Moderate
Placement
0.62
Possible
Fan-In
0.55
Possible
Fan-Out
0.48
Low
Cycling
0.22
Unlikely
FINTRAC Special Projects
Underground Banking
3 of 7 red flags matched
● Multiple cash deposits just under $10K
Fentanyl
1 of 9 red flags matched
○ Cross-border wire to known corridor
Trade-Based ML
0 of 6 red flags matched
○ No match
Human Trafficking
0 of 8 red flags matched
○ No match
Red Flag Indicators
RFIL-
Multiple cash deposits just below $10,000 reporting threshold (7 deposits of $9,200-$9,900 in 30 days)Source: CWAE-MMD + TM Rule · Confidence: High
-
Transaction volume inconsistent with declared income — 7.5x declared annual income in 3 monthsSource: LRP Attribution · Confidence: High
-
Funds received from 6 unique sources then rapidly dispersed to 3 destinations within 48hSource: GNN Pattern · Confidence: Medium
-
Cross-border wire transfers to high-risk jurisdiction (Hong Kong) — 4 wires in 3 months, no prior historySource: CWAE-MMD · Confidence: Medium
-
Account dormant for 14 months prior to sudden activity burstSource: Behavioral · Confidence: Low
Investigation Narrative
LLM-CoTStep 1: Overview
Sarah Chen (RC-2024-08821) is a retail client in Vancouver, BC, employed as a restaurant server with a declared annual income of $38,000. Over the past 3 months, her account has processed $28,700 in transactions — approximately 7.5x her expected monthly volume of $4,200.
Step 2: Structural Analysis
The transaction network reveals a gather-scatter pattern: 6 unique source accounts funnel funds into Chen's account, which then disperses to 3 destination accounts via wire transfers. Two of the source accounts (A4, A5) show direct connections to other source accounts, suggesting coordination.
Step 3: Pattern Matching
This activity is consistent with layering (confidence: 0.91) and structuring (confidence: 0.78). The cash deposits are structured just below the $10,000 CTR threshold. The gather-scatter topology is a hallmark of smurfing networks where multiple individuals deposit cash that is consolidated and moved internationally.
Step 4: Recommendation
Escalate for L2 investigation. Key areas: (1) verify source of funds from 6 counterparties, (2) confirm employment income against declared, (3) investigate the 4 wires to Hong Kong — purpose and relationship to beneficiary. Consider STR filing for structuring if source-of-funds cannot be established.
Add-On Model Signals
PLUG-INTM Rule Engine
—
Not connected
Isolation Forest
—
Not connected
Sanctions Screen
—
Not connected
Custom Model
+
Add plug-in