RISK CALIBRATION LAB
Calibrate with evidence. Deploy with control.
Test proposed thresholds, weights, and parameter configurations against historical trading activity before they affect production. Compare results with the current authorized baseline, understand the impact on alert and entity risk, and move approved changes through a governed deployment process.
TEST→COMPARE→APPROVE→DEPLOY
TEST
Propose & Test Changes
Adjust thresholds, weights, and parameters in a safe, governed environment.
COMPARE
Measure Impact
Compare results to the current production baseline using historical replay.
APPROVE
Governed Approval
Independent review and approval ensure change integrity and control.
DEPLOY
Control Production
Deploy approved packages as the new authorized production version.
THE CONTROLLED CALIBRATION WORKFLOW
Seven governed steps from proposal to production.
1. Select Population
Define the trading context and historical window.
→
2. Load Baseline
Bring in the current authorized production configuration.
→
3. Adjust & Propose
Create a proposed parameter package or scenario.
→
4. Run Historical Replay
Replay historical activity to measure analytical impact.
→
5. Compare Impact
Evaluate differences vs. production across key metrics.
→
6. Submit for Approval
Route the proposed package through independent review.
→
7. Deploy Approved
Approved package becomes the new authorized version.
WHAT THIS MEANS
More meaningful alerts with higher risk coverage
Better focus for reviewers and investigators
Stronger risk detection without unnecessary noise
Data-driven decisions you can defend
PLATFORM MAINTENANCE & GOVERNANCE
Calibration Lab — always-on governance.
Keeps the platform accurate, reliable, and defensible. It supports the engine — it is not part of the daily investigation flow.
LEVEL 3 — PARAMETERS (MARKING THE CLOSE)
PARAMETER WEIGHT TOTAL — 100% VALID
{{ p.name }}
MODIFIED
WEIGHT{{ p.wLabel }}
AUTO-REBALANCE OTHERS
THRESHOLD{{ p.tLabel }}
BREACH SENSITIVITY
ALERT RISK SCORE — LIVE
{{ calibScore }}
{{ calibBand }}
WEIGHTED CONTRIBUTION
{{ b.name }}{{ b.val }}
ALERTS BREACHING THRESHOLD
{{ calibTotalAlerts }}
{{ a.name }}{{ a.count }}
Mechanism demonstration on synthetic data. Weights auto-rebalance to 100%; raising a threshold reduces which activity breaches, and the weighted score recomputes live.
Model tuning
Adjust weights and parameters to reflect risk priorities.
Threshold testing
Test thresholds and scenarios to manage sensitivity.
Score distribution monitoring
Monitor score health, drift, and stability.
Scenario analysis
Evaluate what-if scenarios and market conditions.
Data quality checks
Validate inputs, completeness, and consistency.
Versioning & audit
Track changes, approve versions, maintain governance.
Outcome: a calibrated, trusted, and continuously improving intelligence platform — more accurate risk detection, fewer false positives, stronger outcomes, regulatory defensibility.
THE MNC
DIFFERENCE
DIFFERENCE
Entity-centric, not alert-centric
Multi-dimensional risk intelligence
Explainable & transparent
Actionable insights
Defensible decisions
Better compliance outcomes