Engagement models

Radius can engage in any combination of the modes outlined below.

Consulting

As described elsewhere on this site (Capabilities, Services) Radius have a number of capabilities related to calibration of credit risk models, and credit risk modelling in general.

Working remotely is easier for calibration projects than for full model builds; the lower dimensionality of the data (and therefore lack of borrower entity attributes) simplifies the logistics of data considerably

In terms of pricing, we can quote fixed estimates for projects once the objectives of an engagement have been set, and some preliminary work has been undertaken. For more complex projects, with multiple objectives, possibly across multiple models, we would quote a daily rate for our team

Deliverables for a consulting engagement depends on the scope, but would typically include:

  • Documentation covering the key project deliverable (e.g. calibration solution)
  • Presentation(s) to key stakeholders
  • Data sets post build
  • Code for reproducing the calibration solution
  • Deployment code
  • Deployment and integration support
01

Documentation covering the key project deliverable (e.g. calibration solution)

02

Presentation(s) to key stakeholders

03

 Data sets post build

03

Data sets post build

04

Code for reproducing the calibration solution

04

Code for reproducing the calibration solution

05

Deployment code

06

Deployment and integration support

Managed Services

In keeping with trends in other industries, we are seeing a sharp increase in the use of managed services for well-defined use cases in technical domains. This is true also of model calibration and ‘devops’ for machine learning (MLOps) in general.

Radius can provide a comprehensive, end-to-end service for the calibration of a model, or multiple models, that would include any/all of the following:

  • Monitoring (at any frequency, including RT)
  • Periodic validation (including documentation)
  • Dashboard provision
  • Query support
  • Pre-emptive re-calibration
  • Default rate modelling and forecasting
  • Trend analysis & forecasting
  • Early warning and anomaly detection
  • Configurable notifications

Some modest level of systems integration would be required for the more advanced configurations where default rate.

Modelling Platform

Radius has been developing a SaaS-based, No Code modelling platform for use by banks to develop a range of regulatory models. This platform will both radically compress development times and increase the overall quality of modelling projects.

This platform will be released later this year following a cycle of external testing.