No good model was ever built on bad data
The quality of data always affects the quality of analytics, irrespective of the amount of data available. However, turning raw material into semi-finished product ready as input for further analytics, does not only involve technical challenges (requiring IT skills). It involves bridging the gap between the IT and the Modelling-context.
We interact between the analytics unit and IT. We do this by establishing a common language between the various stakeholders. In addition we identify and unlock relevant data sources into a dedicated platform. Subsequently we preprocess data to create an appropriate data-analytics platform. This involves: linking data and data cleansing (to enhance its quality). Such is easier said than done. Execution is very demanding, time consuming and requires smart people that are willing to make dirty hands.
Our data related services are highly relevant for any business area in which data analytics are applied. In case of more complex questions to answer with data analytics, data challenges will only increase making our services of even more value added. In terms of business areas we focus on financial modelling, financial crime analytics and data analytics (a broader context, also outside the financial sector).
We have vast experience in data preparation within large banks aiming at advanced financial risk modelling. Specific challenges that we have dealt with are: unclear data-field definitions, combining different source systems, historical changes in processes (e.g. in work out of loans).
Recalibration of credit risk models after bank-wide new definition of default
Due to a bank-wide new definition of default the credit risk model landscape at a major bank needed recalibration
For equity funds, a bond- ,an asset-backed security fund and a real estate fund
Validation Project Finance PD model
Support a bank in the validation of their Project Finance PD model