Credit risk moves daily — but agency ratings don't. The gap between a firm's actual creditworthiness and its last agency rating can persist for months, leaving portfolios exposed to risk that static scores fail to capture.
Criat Probability of Default (PD) quantifies a firm's default likelihood from 1 month to 5 years; PD-Implied Rating (PDiR) maps it to a letter-grade scale directly comparable to agency ratings — but updated daily.
Silicon Valley Bankheld investment-grade ratings from both Moody's and S&P until the day it collapsed.
PD data flows into your existing risk infrastructure — whether you analyze interactively, integrate via feeds, or access natively within partner platforms.
Criat's Probability of Default is produced by the Forward Intensity Model (FIM) — a third-generation credit risk model built on econometrics and refined with machine learning.
Dynamic, forward-looking. Built for all firm types and markets.
Forward Intensity Model is a single framework that produces daily, forward-looking default probabilities — across all horizons, firm types, and markets.
Each model draws from up to four risk factor dimensions — macro-financial conditions, firm-specific financial ratios, equity-based factors, and bond-based factors — combined based on data availability and market characteristics.
The foundation. 16 risk factors — level and trend used jointly to capture momentum and directional shifts. Calibrated separately by region and sector.
Extends the Public Firm framework to unlisted companies. Equity-based factors are proxied, producing PD directly comparable to public firms.
Purpose-built for China's onshore and offshore bond markets, where equity data alone is less informative and local macro dynamics diverge from global patterns.
Trained on a meticulously curated default database. Default definition aligned with Big 3 rating agencies — covering bankruptcy, missed payments, and distressed exchanges.
Calibrated across North America, Europe, APAC Developed, Emerging Markets, China, and India — separately for Corporate and Financial sectors.
Criat's in-sample and out-of-sample Accuracy Ratios (AR) are consistent — validating that the model generalizes beyond its training data without overfitting.