The author

Mark Vickers

Head of IFRS 9

View profile
News & Views / How Horizon can help with benchmarking your IFRS 9 outputs
27 June 2017

How Horizon can help with benchmarking your IFRS 9 outputs

As organisations start to see light at the end of the tunnel on IFRS 9 modelling, Jaywing is pleased to offer an independent validation service that uses our IFRS 9 software Horizon to compare your provisioning outputs using our methodology, which has been deployed with over a dozen of the UK’s lenders.

Our IFRS 9 annual validation can be used to get outputs from an alternative modelling methodology, using our Exogenous-Maturity-Vintage approach, either as an annual comparison of model performance or as a one-off review, whenever it is required.

Taking your summarised portfolio performance data as inputs, and coupled with your own choice of economic scenarios, the validation will act as a sense check on the models that you have deployed – enabling comparison with a methodology that is widely used across the industry. 

The validation is offered as a consulting service, with our experts working with you to source the relevant data and process it in our IFRS 9 modelling tool Horizon, producing outputs and comparisons to your own model outputs.

In addition to validation, Horizon can help in a number of other regards. It can help in stress testing exercises to assess impact on a range of metrics, including provision.  Alternatively, it can be used to undertake IFRS 9 modelling on any portfolio on which you have not yet been able to complete your own development, or as a backstop position in case there is any risk to the timing of your parallel run or live deployment.

Horizon uses advanced machine learning techniques to quickly develop accurate, robust models, and ensures that its models always behave ‘sensibly’ in response to changing economic scenarios and the data within your files. It always runs from the latest data position, meaning that you always have the most up to date model possible, and aren’t constrained by ageing models that were built on old data samples.