|
|
| The team calls upon a wide range of analytic methods and model types to solve risk-based problems. |
 |
At the heart of our approach is the multi-faceted discipline of Operational Research (OR), a branch of applied mathematics that integrates elements of statistical analysis, optimisation and simulation. Being conversant in multiple, complementary techniques maximises the chance of building predictive models.
In addition to these core elements, we have developed other capabilities that have great modelling utility for the problems we encounter. This includes various Bayesian methods for where the client has issues with data—or where a degree of approximation or expert input is required. We have also utilised a number of non-parametric and numerical methods derived from algorithms. Numerical methods have value where problems have complex, intractable structures or exhibit high levels of interdependency between factors.
In the course of recent work, we have also developed entire methodologies (i.e. collections of related techniques and processes) from first principles to solve problems specific to the project.
Despite the emphasis on technical methods here, we are not in the practice of forcing complex procedures onto a problem. With each project, we look for the simplest, most efficient and transparent route possible to develop solutions.
We also provide comprehensive documentation of the methodology. Our work is communicated in plain language, as well as technical terms, such that clients can follow our assumptions and verify our results. Client briefings and workshops are liberally spread through our working process: this helps to clarify our assumptions and choice of methods.
This technical foundation complements and informs our experience in banking, credit risk, asset finance, insurance and other areas of financial services.
|
| Core capabilities |
|
| Credit risk modelling within the Basel II context |
| • |
Advanced credit-risk modelling methodologies (such as for the development of PD, LGD and EAD models, and the calibration of rating scales) |
| • |
Non-parametric techniques for problematic data sets |
The company has experience in developing innovative approaches to support Basel II requirements—especially for so-called ‘low default portfolios’ and other situations where mainstream statistical methods fail.
A number of portfolios and sectors have been addressed by the team: from retail portfolios with large datasets to small, specialised portfolios with fragmented datasets. White Kite has consequently developed a range of techniques for mathematically integrating expert opinion into statistical models. |
|
| Statistical Modelling |
| • |
Multivariate statistical analysis and exploration of data (for classification, prediction) |
| • |
Predictive modelling for forecasting or general risk modelling |
| • |
Bayesian modelling for integrating expert opinion into traditional models |
The most common methods for analysing business problems and building models are statistical (descriptive and inferential). White Kite has broad experience in multivariate statistical modelling, including all mainstream forms of regression modelling for predictive applications. |
|
| Optimisation |
| • |
Mathematical programming (linear, integer, mixed) |
| • |
Non-linear optimisation |
Optimisation methods suit scheduling, pricing and asset allocation (portfolio) problems. White Kite has experience in using advanced linear programming methods to create and deploy enterprise-level solutions, such as might be applied to whole portfolios and at high frequency. |
|
| Simulation-based modelling |
| • |
Monte Carlo simulation and sampling (for both financial and operational applications) |
| • |
Discrete event simulation for operational problems (such as modelling the behaviour of physical networks and workflows) |
Simulation methods are very powerful for analysing problems where there is (1) uncertainty and/or variability in the key factors as well as (2) a high level of complexity, possibly in the form of dependency between factors.
As indicated below, we also have considerable experience in using Monte Carlo simulation for modelling portfolios of assets and designing scenarios for stress-testing. |
|
| Portfolio Risk Modelling |
| • |
Portfolio risk analysis |
| • |
Stress testing and sensitivity analyses |
We have experience in analysing various cash-flow based portfolios for risk of loss, using mainstream VaR (‘value at risk’) and other distribution-based methods. White Kite is able to build many different types of portfolio risk model including portfolio models of projects, and transportation assets, as opposed to standard models of liquid market securities and derivatives.
Recent work has also included the adaptation of a traditional market risk framework to solve the policy component of a Basel II problem. |
|
| Forecasting |
| • |
Econometric analysis and modelling of time series |
| • |
Non-parametric methods for forecasting short-run series |
Our experience in econometric modelling includes the development of theoretical models for long-run forecasts; ‘atheoretical’ models for shorter horizons; and models based on stochastic differential equations.
For forecasting ultra-short term effects, we have experience in the use of advanced algorithmic and non-traditional methods, such as those used in pattern recognition applications.
|
| TOP |
|
| White Kite: +44 (0) 207 234 9470 |
| Email: risk@whitekite.com |
|
