PREDITTA Lenses/Advanced Risk Solutions
financial statements, payment behaviour, bureau data and public market data converge into a rating, a probability of default and a suggested limit.
Every number carries the methodology that produced it, and every company is placed against its peer group rather than an absolute threshold.
Built for the transition to expected loss models
The Brazilian market is moving to expected loss models under Resolution 4.966 and IFRS 9. Preditta delivers the analytical components that underpin that transition: a probability of default model trained on payment behaviour, and the sector benchmarking that gives every grade its meaning.
The Platform
Analytical consulting and a SaaS platform, carrying decades of SIACorp credit automation heritage.
The expert grade walks 34 indicators positioned against a peer group. The statistical model estimates a probability of default. Both readings sit on the same company, and disagreement between them is information.
Balance sheet, income statement and supporting filings arrive as PDF, XLSX or XML. Extraction is AI-assisted, account mapping stays visible, and review is human before any calculation runs.
Sector reports built on open sources — CVM-listed companies and the Central Bank credit registry — to place each company inside its sector rather than judging it in isolation.
Modules
Thirty-four indicators across seven families — liquidity, profitability, leverage, asset composition, working capital, cash flow and operational leverage — converted into percentiles against the peer group and into a grade from AA to D.
A probability of default model trained to predict the following year's outcome, validated company by company, and published alongside its discrimination, calibration and variable importance metrics.
Assessment of thin-file records — small businesses with no statements available — from a credit bureau query and the economic activity group alone.
Payment capacity, a solvency ceiling and an expected loss factor combined into a proposed limit, with every parameter adjustable and shown next to the result.
Receivables aging, average collection period and value-weighted delay, portfolio concentration, and per-client payment prediction.
Questions in plain language translated into queries over a fixed data catalog, held to read-only by independent validation layers and recorded in an audit trail.
Our DNA
28 years of SIACorp credit automation combined with machine learning applied to risk. We inherit decades of credit automation and join it with state-of-the-art AI to deliver end-to-end risk assessment.
No grade without an explanation
Every grade in the platform has a page describing how it was produced: which indicators feed it, how they become percentiles, and how the bands are set. A risk number that cannot be audited cannot carry a credit decision.
Get Started