Application Fraud Modelling
We build the models that identify fraudulent applications at the point of application, or validate the models you run.
Banking
Fraud
Detection at two points: the fraudulent application at the door, and the fraudulent behaviour on an account you already hold.
Fraud Deliverables
Behavioural Fraud Modelling
We build the models that identify fraud after origination, reading the transaction and account behaviour that precedes it, with alerting into your investigation workflow.
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
-
Fraud models are judged on what happens after the alert, so ask any bidder what their output does to your investigation team’s workload. At Gini we build application, transaction and identity fraud models, tune anti-money-laundering alerting to cut the false positives that consume investigator time, and use graph analytics to surface organised fraud that account-level scoring cannot see.
-
Yes. Gini tests detection rate at your operating false positive rate, stability over time, and whether the features still carry signal now that the fraud has moved. Fraud models decay faster than credit models because the people they detect adapt, so validation looks hard at how recently the model was refitted and what alert volume has done since.
-
Usually, yes, and the gain often comes from tuning thresholds and consolidating overlapping rules rather than from a new model. At Gini we measure each rule’s contribution to genuine cases against the alerts it generates, then retire or retune the rules that produce volume without outcomes. Every change is evidenced against historical cases before it goes live.
-
The graph Gini builds finds the relationships that account-level scoring is blind to: shared devices, addresses, bank details and contact points linking applications that look unremarkable one at a time. Organised fraud is a network before it is a loss, and clusters surface earlier in the graph than they do in any individual score.