About Radius

We are modellers, engineers and risk professionals with extensive experience across multiple domains including aerospace, insurance, e-commerce, and financial services.

Within Risk specifically, we have accumulated deep analytics experience across multiple Tier 1 banks, building and calibrating models to high standards for high-materiality portfolios under the AIRB rules. Our experience has been built primarily around the development and calibration of capital models (PD, LGD and EAD models), and methodology development for same. However, we have been fortunate enough to work on other problems including VaR models for leasing portfolios; early warning systems for currency crises; advanced credit cycle frameworks; and policy simulators.

In terms of project delivery, we have built models ourselves; managed modelling teams as part of a wider regulatory programme; and provided advisory services related to modelling strategies, advanced methodologies, and strategic modelling architectures.

A common theme running throughout our work has been the design and development of predictive models for complex use cases. In the course of this work, we have designed new methodological frameworks from first principles and adapted existing ones. An example of this is the work we have done on methodologies for low default portfolios, including the synthesis of subjective expert opinion into the modelling process.

The diversity of our work has led us to draw on a broad range of methodologies and techniques:

  • statistical model-building for classification, event prediction, asset valuation and forecasting
  • advanced time series methods from econometrics, and signal-processing
  • optimisation frameworks (including tailored metaheuristics) for parameter estimation, model selection, and control
  • generalised risk and probabilistic modelling using Monte Carlo methods
  • Bayesian statistics for low-data environments and hierarchical modelling
  • formal logic for policy analysis
  • a broad palette of machine learning methods for things such as early warning systems and trading models

In addition to modelling work, one of the Founders established and ran a specialised R&D unit within a large European bank. The mission of this unit (‘Strategic Analytics’) was to design and build a cluster of inter-locking modelling utilities, components, and execution environments that could accelerate the development of models and improve key operational parameters of the model life cycle more generally. With people in Cambridge, London, New York and Kraków, the unit worked on a number of strategic projects across wholesale credit risk, traded credit risk, market risk, country risk, and financial crime.

Our long-term strategic focus is on the design and development of cloud-based architectures for the acceleration and robustification of modelling workflows. We are on the cusp of launching an advanced modelling platform that supports multi-stakeholder workflows and treats model-building as a manufacturing problem.

Within risk environments, we have extensive experience across multiple Tier 1 banks and large corporate environments including:

  • Performing model builds and designing advanced calibration solutions;
  • Performing model builds and designing advanced calibration solutions;
  • Deep knowledge of low default portfolios;
  • We have established a range of methodologies for the synthesis of subjective expert opinion into the modelling process;
  • Early warning and policy simulators;
  • Data analysis and model building using remote sensing;
  • Cash flow modelling in leasing, and advanced simulation-based models for company valuation.

Our Mission

We build cloud-native software for the accelerated development of advanced predictive models, in demanding environments, to exacting standards.