Accurate*

AccuRate* is a set of tools for calibration, validation and automated selection of PD models, which assigns corresponding credit scores and alphanumeric rating grades to borrowers and individual claims, based on their PD estimates. It provides state of art internal rating model fully aligned with regulatory requirements for transition from standardized to internal ratings based approach to calculating risk weighted assets under Basel framework. Smart brute force algorithm has been implemented to allow for globally optimal model search based on model predictive power.

AccuRate* is developed in a multi-layered architecture. Its components include:

  • Layer for import and validation of data about private individuals and legal entities
  • Layer for model calibration with predefined parameters without user interaction
  • Layer for model calibration with predefined parameters with full user interaction

AccuRate* employs logistic regression techniques for calibrating PD models, which are considered an industry standard and are recognized as the best practice, both by banks and their regulators. The result of the estimation is a PD model which is then subject to thorough validation, using in-sample and out-of-sample statistical tests, as well as by performing out-of-time model performance tests. The proceeding automated selection of optimal model specification is implemented through information criteria. Lastly, borrowers and exposures are assigned credit scores and mapped onto alphanumeric rating scale, using various techniques of statistical clustering.

AccuRate* has the following functionalities:

  • Calibration of PD models, based on both quantitative and qualitative data (soft facts);
  • Validation of PD models, using in-sample, out-of-sample and out-of-time statistical tests;
  • Automated selection of optimal PD models among several candidate-models;
  • Assignment of credit scores to individual borrowers in sample, based on estimated PD;
  • Mapping borrowers and exposures onto alphanumeric rating scale;
  • Generating various predefined reports.

AccuRate* utilizes advanced methodologies for model calibration and validation in order to completely bypass common problems of standard logit models: multicollinearity, extreme factor values, missing values, and non-linear relationship between log odds and factor itself. These are overcome using AccuRate* by applying various data transformations and Firth’s bias reducing regression.

AccuRate* is well ahead of its peers both in terms of complexity and speed of calculations.

Key features of AccuRate* are the following:

  • Built-in tool which enables users to create custom factors based off of raw data;
  • Allows for expert opinion in factor selection process, in order to best capture specific local market conditions;
  • Multiple competitive models, which have been successfully validated, are presented and users may opt in for one of them;
  • Immense flexibility for parameter selection is enabled to including selection of multiple stratification methods – by sector, region, period etc – in order to improve model performance;
  • Highly intuitive user interface which provides progress bar monitoring and reversibility to previous steps option;
  • AccuRate* has its own relational database and built-in tool for data import (highly intuitive, fast and simple for use);
  • AccuRate* performs automated out-of-sample and out-of-time validation of models, which is prerequisite for transition to internal ratings based approach and for internal ratings audit;
  • AccuRate* creates rating categories based on calibrated model by applying state of art clustering technique and calculates optimal number of rating categories based on any given model;
  • AccuRate* enables archiving of relevant steps, intra-steps, results, calibration and validation of data, and allows for export into Microsoft Excel.

AccuRate* comes with accompanying internal ratings model methodology where detailed introduction to statistical methods applied in modeling process is provided.