13.0 Simulated Alert Outputs with Explainability

This page provides simulated alert outputs based on Kor.ai’s Bayesian model and explainability design. Each alert contains a scored posterior (MarketAbuseLikelihood), a structured evidence breakdown (ExplainabilityModule), and user-facing rationale to support review and triage workflows.

---

🧪 Example 1: Insider Dealing Scenario

Alert Metadata

• Trader ID: T567
• Date: 2025-05-17
• Desk: MENA Equities
• Asset: Commodity Producer Equity

Posterior Output

• MarketAbuseLikelihood: High

ExplainabilityModule

{
  "AccessToMNPI": {
    "value": "Yes",
    "score_impact": 0.35,
    "confidence": "High",
    "input_description": "Trader accessed internal sales report 1h before trade."
  },
  "TradeDirectionAligned": {
    "value": "Yes",
    "score_impact": 0.20,
    "confidence": "Medium",
    "input_description": "Trade benefited from forecasted downgrade."
  },
  "TimingProximity": {
    "value": "<1h",
    "score_impact": 0.15,
    "confidence": "High",
    "input_description": "Trade placed 37 mins after access."
  },
  "PNLSpike": {
    "value": "Extreme",
    "score_impact": 0.20,
    "confidence": "High",
    "input_description": "Largest intraday PnL in 6 months."
  },
  "KnownRiskProfile": {
    "value": "PreviouslyFlagged",
    "score_impact": 0.10,
    "confidence": "Medium",
    "input_description": "Previously reviewed for similar trade behavior."
  }
}

---

🧪 Example 2: Spoofing Pattern

Alert Metadata

• Trader ID: T233
• Date: 2025-06-01
• Desk: Oil Derivatives
• Asset: Crude Futures

Posterior Output

• MarketAbuseLikelihood: Medium

ExplainabilityModule

{
  "OrderPlacementRate": {
    "value": "High",
    "score_impact": 0.25,
    "confidence": "High",
    "input_description": "112 orders placed in 5 mins."
  },
  "CancelRateRatio": {
    "value": ">90%",
    "score_impact": 0.30,
    "confidence": "High",
    "input_description": "104 of 112 orders cancelled within 30 sec."
  },
  "BookPressure": {
    "value": "AskPressure",
    "score_impact": 0.15,
    "confidence": "Medium",
    "input_description": "Over 60% imbalance during quote cycle."
  },
  "LayeringPattern": {
    "value": "Moderate",
    "score_impact": 0.20,
    "confidence": "Medium",
    "input_description": "Detected layer pattern over 3 quote cycles."
  },
  "ClusteredBehavior": {
    "value": "Similar",
    "score_impact": 0.10,
    "confidence": "Low",
    "input_description": "Similar behavior flagged last quarter."
  }
}

---

🧪 Example 3: Front Running

Alert Metadata

• Trader ID: T998
• Date: 2025-06-11
• Desk: Structured Rates
• Asset: EUR Gov Bonds

Posterior Output

• MarketAbuseLikelihood: High

ExplainabilityModule

{
  "ClientOrderVisibility": {
    "value": "Full",
    "score_impact": 0.30,
    "confidence": "High",
    "input_description": "Trader assigned to internal client execution desk."
  },
  "PreTradeBehavior": {
    "value": "SignificantFrontRun",
    "score_impact": 0.25,
    "confidence": "High",
    "input_description": "Bought 3 mins before client ticket."
  },
  "TradeDirectionAligned": {
    "value": "Yes",
    "score_impact": 0.20,
    "confidence": "Medium",
    "input_description": "Front-run trade aligned with client direction."
  },
  "PNLSpike": {
    "value": "Mild",
    "score_impact": 0.10,
    "confidence": "Medium",
    "input_description": "Trade resulted in notable profit."
  },
  "KYCRiskRating": {
    "value": "High",
    "score_impact": 0.15,
    "confidence": "High",
    "input_description": "Trader flagged in KYC profile as enhanced risk."
  }
}

---

These outputs can be fed into the Kor.ai UI alert detail view and optionally piped into downstream audit or STOR workflows. Score impact and rationale mapping supports analyst prioritization, review audit trails, and explainability for compliance stakeholders.

Next: Link to test payloads and UI rendering logic in the Alert Viewer module.