Most machine learning models used for underwriting get stuck in the lab. Our models get results. ZestFinance helps leading financial institutions to harness the power of machine learning. Our software, called ZAML, provides access to the most comprehensive suite of machine learning tools built specifically for underwriting and is backed by a team of world-class coders, modelers, and technologists who are available to support you every step along the way. We make implementation and compliance easy and enable you to get your model into production fast. Our customers typically achieve a 15% increase in approval rates or a 30% reduction in charge-offs.
Take advantage of the full power of machine learning to turbo-charge your results with the assurance that every underwriting decision – on every loan – can be easily explained to the members of your risk
committee, board, or regulators.
The ZAML suite is comprised of six modules:
ZAML tools give you the freedom and flexibility to build powerful models in your own environment. Our tools are built in R and Python so they work in all conventional modeling environments. Or, our team can provide a build environment for you. ZAML tools help boost predictive power by harnessing more from your existing internal customer information and credit bureau data and then synthesizing from thousands of variables into a single credit model. No additional alternative or third-party data is ever required.
ZAML tools help you quantify model risk and performance and determine the safety constraints for your ML models. ZAML tools clearly identify bias in your model so you can make the necessary adjustments to comply with the Equal Credit Opportunity Act (ECOA). You can also compare newly approved applicants with prior applicants to understand the risks of adopting new models.
ZAML tools allow you to rapidly iterate between the build and test stages so your model can comply with all business, legal, and compliance requirements. That lets you make adjustments to your model faster – so you can get out of the lab in just a few months. ZAML tools allow you to conveniently conduct champion-challenger analysis so you can more accurately project the potential improvement in approval rates, loan volume, and charge-off rates.
You can run a compliant model where you’d like - on premises, in the cloud, in containers, on platforms like Microsoft, Amazon, or Google, on open source environments or any other environment that your IT team prefers. ZAML tools accelerate your deployment time by allowing the business teams to directly operate their own models with no recoding. You can refit and deploy new models as markets change easily. No proprietary run-time engine is required.
ZAML tools continuously evaluate model results and alert you when the applied population is inconsistent with the expectations and assumptions built into your model.This allows you to identify and mitigate potential risks before they happen. ZAML tools continuously monitor your applicant pools to identify statistically relevant applicant pool changes. That empowers you to determine whether you need to refit your data to the model if the underlying borrower applicant shift.
ZAML tools open the black box so you can explain your machine learning model’s credit decisions for the entire loan portfolio – and every single applicant, including denied borrowers, as required by the Fair Credit Reporting Act (FCRA) and the EU’s GDPR. ZAML tools automatically produce any documentation you need for model development and deployment, saving you time and eliminating errors. ZAML can produce model risk management reports, fair lending and economic impact analysis, and adverse action safety alerts to make sure you are satisfying all regulatory requirements.
Our customers who switch to transparent ML underwriting models see a 15% typical increase in approval rates with no added risk, or choose to capture the gains in reducing risk, where they see a typical 30% reduction in charge-off rates.
ZAML tools help you:
Take better advantage of the data you already have.
ZAML tools allow your model to combine thousands of variables for deeper insights. They are platform-agnostic and work on-premise or in the cloud.
Find more good customers and fewer bad ones.
ZAML tools allow you to add more creditworthy borrowers while maintaining your risk profile. That helps you more sustainably and profitably grow your loan book.
Run your model safely – and with confidence.
ZAML tools make compliance easy. They automate risk management reports, fair lending and economic impact analysis, adverse action reason codes, and help you implement comprehensive model safety monitors – so you can focus on great modeling.