Prediction
Forecast evaluation under asymmetric loss
Why accuracy alone is insufficient when false negatives, false positives, limits, and interventions carry different costs.
Research noteResearch
Modeling methods and validation patterns for risk teams.
Representative research notes connect quantitative methods to operating limits, evidence, and human decisions.
4 selected resourcesPractical, inspectable, and clearly scoped
No fabricated customer names, performance claims, or unavailable downloads.
Selected content
Designed for evaluation, validation, and delivery.
These public summaries explain FiboRisk methods and product patterns. Customer-specific evidence remains inside the agreed private environment.
Anomaly detection
Graph anomalies with investigator feedback
How connected entities, sequences, and case disposition can improve prioritization without hiding context.
Research noteScoring
Calibration stability across segments
Testing probability calibration, grade migration, thresholds, and fairness across operating segments.
Research noteSimulation
Sensitivity and reverse-stress design
Comparing deterministic, probabilistic, sensitivity, adverse, and reverse-stress methods.
Research noteNeed the customer-specific version?