Five-Stage Architecture
Stage 01
MAP
Stage 02
SCORE
Stage 03
DECIDE
Stage 04
MONITOR
Stage 05
LEARN
STAGE 01
MAP
Criterica Underwrite
Map the matter, normalize messy inputs into features, and generate the first full prediction set with stated confidence.
Live
Production intelligence models · calibrated across litigation finance, insurance, law, and courts
Live
Pre-commitment scoring · validated outcome probability before capital is deployed, trained on real court records
Live
Judicial intelligence · judge-level decision history, venue bias, assignment probability at circuit, district, and individual level
Live
Duration, value, and exposure modeling · filing to resolution timing, low / base / high bands, red-flag detection
Roadmap
Data-confidence and missing-data impact scoring · which inputs most improve accuracy
STAGE 02
SCORE
Criterica Underwrite
Run the full prediction engine. Output a calibrated probability set with confidence intervals across outcome, duration, and value.
Live
Outcome probability scoring · calibrated win/settle/dismiss likelihoods with jurisdiction-specific baselines
Live
Value band modeling · low / base / high expected recovery with historical distribution anchors
Live
Confidence-weighted composite score · single decision-ready signal from multi-model ensemble
Roadmap
Adversarial feature injection · stress-test the score against opposing counsel patterns and judge variance
STAGE 03
DECIDE
Capital Partners
Convert prediction into a priced decision: fund, decline, reserve, settle, or assign, with the expected impact of each path.
Live
Capital deployed across funded advances · non-recourse, intelligence-underwritten, 24-hour standard decisions
Live
Settlement window and timing analysis · when value crystallizes, when to open talks
Live
State-by-state regulatory posture · practices aligned to ILFA and ALFA codes · model-governed underwriting discipline
Live
Underwriting intelligence for funders · pre-settlement, law firm lending, litigation finance, medical lien receivables, portfolio finance, special situations
Roadmap
ABS jurisdiction structures · AZ, UK, AU non-lawyer ownership frameworks
Roadmap
Insurance reserve and claims gates · model coverage live, decision product in build
STAGE 04
MONITOR
Living Score
The living score — the model watches every matter in the portfolio and flags when something changes.
Live
Portfolio-level score drift detection · matter conditions change, the score updates
Live
Settlement signal monitoring · timing and value crystallization alerts
Live
Judicial assignment tracking · new judge assignment triggers venue re-score
Roadmap
Real-time docket event ingestion · filing events trigger automatic re-scoring
Roadmap
Concentration risk flags · portfolio-level exposure aggregation across venue, judge, and case type
STAGE 05
LEARN
Criterica Ledger
Capture the actual outcome, value, and duration as structured training labels. Every resolved matter makes the next prediction better.
Live
Real court records as the training substrate · real judicial outcomes, not synthetic distributions
Live
Outcome capture pipeline · actual result submitted against the original prediction
Live
Published methodology and governance · stated model provenance and limitations for regulated buyers
Roadmap
Calibration reporting · predicted vs. actual error tracking, published per model family
The Underwriting Equation
Expected Value×
Probability×
Duration×
Confidence×
Cost of Capital
Duration is capital velocity. A 70% settlement in 11 months can outperform a full win in 36. Criterica prices both dimensions at every stage.
Closed Loop
ACQUIRE
IMPROVE
PRICE
DEPLOY
GENERATE
STRENGTHEN
