Application Scorecards
We build the models that decide which applications you approve, and evidence their accuracy, stability and fairness.
Origination
Strategy
How an application becomes a decision: who you approve, on what terms and at what price.
Origination Strategy Deliverables
Affordability Modelling
We build the assessment of whether a customer can service the repayments.
Profitability and Pricing Modelling
We calculate the profit a customer generates across the whole relationship, and work back from it to the price the product has to carry.
Lending Strategy
We design and test the scores, policies and rules that turn an application into a decision, covering credit risk, pricing, affordability, identity and fraud checks, account management and collections.
Banking Case Studies
Banking
Independent validation of a Pillar 2A concentration add-on
A European bank needed to know that the capital it holds against concentrated lending would stand up to scrutiny, so it asked us for an independent view of the model behind the number.
Banking
Redeveloping every IFRS 9 model across two portfolios
A UK lender needed every model in its IFRS 9 expected credit loss suite rebuilt, at the point when the team that had built them was no longer there.
Frequently Asked Questions
-
Look for a partner who owns the whole chain from scorecard to decision engine, because most origination strategies fail at the joins rather than in the statistics. At Gini we build application scorecards, affordability and income verification models, and the cut-off and referral strategy that sits on top of them. The strategy is implemented in the decision engine you already run, and champion-challenger testing shows what the change earned rather than what it was forecast to earn.
-
Yes, and a diagnostic is usually the cheaper first step. Gini tests the incumbent on discrimination, calibration and stability by segment, then builds an independent challenger from your raw data so you can see where the two disagree and on which applications. Where the existing model still ranks risk well and the problem sits in the cut-offs or the override rules, we say so and leave the model alone.
-
By designing the test before the change goes live. At Gini we set the control and challenger populations, the metric each has to move and the period over which the comparison holds, then report acceptance rate, expected loss and margin side by side. Without that design fixed in advance, a strategy change and a shift in the applicant mix look identical after the fact.
-
Affordability is modelled on your own outcomes rather than on a generic expenditure benchmark, and the assumptions behind each threshold are documented so a supervisor can follow them. Gini tests for disparate outcomes across the characteristics the FCA expects you to monitor, and where a variable improves discrimination but cannot be justified to a declined customer, we exclude it before the model is built.