01
Vintage analysis for credit policy
Compare cohorts before and after a policy or pricing change.
Financial Services / Lending Analytics
Lending analytics built from loan-level data, not month-end totals
A question like 'are this year's loans performing worse?' cannot be answered from balance totals. Agnotic Technologies builds lending analytics on monthly loan-level performance snapshots, so credit and finance teams get vintage curves, roll rates, early default and prepayment views, plus a clean input set for the CECL model.
Vintages
Loans grouped by origination month
Roll rates
Status moves month to month
CECL inputs
Loan-level data for finance
Trusted by global innovators
Industry Overview
What is lending analytics?
Lending analytics is the measurement of how loans perform after they are made: which origination cohorts lose more, how accounts move between delinquency states, how many default early and how fast balances prepay. Agnotic Technologies builds lending analytics on loan-level monthly snapshots and hands finance a governed data set for expected credit loss models.
Industry Challenges
Why can't most lenders answer vintage questions?
The data usually exists, but not at the grain these methods need.
The warehouse keeps only each loan's current status, so last year's delinquency state is overwritten and roll rates cannot be rebuilt.
Charge-offs, recoveries and modifications live in servicing notes rather than as dated events, so loss curves miss them or count them twice.
Loans migrated from an earlier system lose their origination date and land in the wrong vintage.
Each team defines early payment default differently, so credit and capital markets report different results for the same loans.
Paid-off loans are dropped from reporting tables, which hides prepayment and makes each cohort look better than it was.
The CECL model is fed a spreadsheet extract nobody can rebuild, so finance cannot show which loans made up last quarter's inputs.
Metric definitions
How is each lending analytics metric defined?
We write these definitions down with your credit and finance teams before building any chart. The formula and the grain must match in every report.
| Metric | Formula in words | Data grain | Source system |
|---|---|---|---|
| Vintage cumulative loss | Net charge-offs to date on loans originated in a period, divided by the original principal of that period's loans, tracked by months on book | Loan by month on book, grouped by origination month | Origination system for cohort and amount; servicing for charge-offs and recoveries |
| Roll rate | Balance or count of loans in one delinquency state this month that sit in the next worse state next month, divided by the starting balance or count | Loan by month, with status at month end | Servicing system monthly snapshot |
| Early payment default | Loans that reach your defined delinquency state within your defined number of payments after the first due date, divided by loans originated | Loan, first payments only | Servicing payment history plus origination date |
| Prepayment speed | Unscheduled principal paid in the month divided by the balance that was scheduled to remain, often annualized | Loan by month, with scheduled and actual principal | Servicing system and amortization schedule |
| CECL reserve inputs | Loan-level balance, amortized cost inputs, risk attributes, contractual term, expected prepayment and historical loss data passed to the finance model | Loan by reporting date | Servicing, origination and general ledger, reconciled |
Our Approach
How should a lender approach lending analytics?
Three choices decide whether the numbers survive a challenge from finance, audit or an investor.
Snapshot every loan every month
Store each loan's status, balance and key attributes at every month end and never overwrite them. Every vintage and roll-rate view is rebuilt from these rows.
Turn servicing activity into dated events
Charge-offs, recoveries, modifications, payoffs and due date changes become events with an effective date, so loss and prepayment curves count each one exactly once.
Reconcile before you analyze
Snapshot balances tie to the general ledger each month before any metric is published, and the CECL input set carries that reconciliation with it.
Capabilities
What do we build for lending analytics?
How to build lending analytics that credit and finance teams can both rely on.
Loan-level snapshot table
One row per loan per month end, including closed and charged-off loans, never overwritten.
Servicing event table
Charge-offs, recoveries, modifications and payoffs with effective dates and amounts, traced to the source system.
Vintage curves
Cumulative loss and delinquency by months on book for each origination cohort, filterable by product, channel and grade.
Roll-rate matrices
Month-to-month transition matrices between delinquency states, by balance and by count.
Early payment default monitor
First-payment performance by channel, partner and policy version, using one agreed definition.
Prepayment analysis
Monthly and annualized prepayment by cohort, rate band and product, from scheduled and actual principal.
CECL input export
A governed loan-level data set for the finance model, with its general ledger reconciliation attached.
Metric definition registry
Each metric's formula, grain, source and owner held under version control, so a definition change is visible.
How It Works
How is the loan-level data set built?
The pipeline we build under lending analytics. Each layer can be rebuilt from the one before it.
Extract
Loan, payment and event data are pulled from origination and servicing systems daily, keeping source keys so every row traces back.
01
1
Conform
Products, statuses and dates are mapped to one definition set, including migrated loans with their true origination date.
02
2
Snapshot
At each month end, one row per loan is written with balance, days past due, status, rate, remaining term and risk attributes.
03
3
Record events
Charge-offs, recoveries, modifications, payoffs and other servicing events are stored with effective dates and amounts.
04
4
Reconcile
Snapshot balances and charge-offs are tied to the general ledger, and a month is published only when it reconciles or the difference is explained.
05
5
Serve
Vintage, roll-rate, early default and prepayment views are built, and the CECL input set is exported with its reconciliation.
06
6
Engineering Stack
Tools behind lending analytics
Chosen to fit the warehouse and BI tools you already run.
Data platform
Cloud data warehouseSQL transformation models under version controlOrchestration with data quality checks
Methods
Cohort and vintage tablesTransition matrices for roll ratesPrepayment from scheduled and actual principal
Delivery
BI dashboards for credit teamsGoverned extracts for finance modelsNotebook access for analysts
Controls
Row-level lineage to source systemsVersion-controlled metric definitionsAccess by roleMonthly reconciliation reports
Measured Impact
How do you know the lending data can be trusted?
Measures of the data itself, not of the loans, that we report to data owners every month.
Completeness
Loans with a month-end row
Active and closed loans present in every month they existed.
GL tie-out
Balance differences
Gaps between snapshot balances and the general ledger, by product.
Definitions
Metrics with one owner
Published metrics that have a written definition and a named owner.
Rebuild time
Month-end refresh
How long a month takes to rebuild after a late correction.
Use Cases
Where do lenders start with lending analytics?
First questions lending analytics answers.
02
Roll-rate forecasting
Transition matrices that feed collections planning and loss forecasts.
03
Early payment default monitoring
Spot fraud or underwriting gaps in a new channel within its first payments.
04
CECL data set
A governed, reconciled loan-level input set for the finance team's model.
05
Funding partner reporting
Loan-level tapes and performance reports for warehouse lenders and investors.
06
Lending analytics for startups
A snapshot table and core vintage views from the first loans, so history exists when partners ask for it.
Integrations
Which systems feed lending analytics?
The sources a loan-level data set is built from, and where it goes.
Cohort and attributes
Loan origination system
Payments and status
Loan servicing system
Reconciliation
General ledger
Events
Collections platform
Finance
CECL model
Dashboards
BI tool
Compliance-First Development Services Backed by Global Standards
We build secure, scalable products designed for privacy, interoperability, and regulatory readiness from day one across every sector we serve.
Service Organization Control 2
Verified controls for security, availability, and confidentiality of enterprise data systems.
Information Security Management
Adhering to the international gold standard for managing information security risks.
Our Edge
Why Global Leaders Choose Us
We combine deep technical expertise with industry-specific knowledge to deliver solutions that aren't just functional, but transformational.
Enterprise-Grade Security
We implement rigorous security protocols and compliance standards (HIPAA, GDPR, SOC2) across all industrial solutions to protect sensitive data.
High-Performance Scaling
Our architectures are built to handle massive data loads and user bases, keeping performance steady whether you're serving ten or ten million.
Accelerated Time-to-Market
Using our suite of internal tools and proven frameworks, we cut setup work so your first release reaches users sooner.
Embedded AI Integration
Beyond simple wrappers, we build deep-learning integrations and predictive analytics directly into the core of your industry-specific workflows.
Engagement Model
How does a lending analytics engagement work?
We start with one loan product and its history. Timing depends on scope; we agree a fixed first release on the scoping call.
01
Scoping call
We review your loan systems, the depth of history, current reports and the questions credit and finance cannot answer today.
02
Definitions workshop
Metric definitions, delinquency states and event types are agreed and written down with named owners.
03
Fixed-scope first release
Monthly snapshots, the event table and vintage and roll-rate views for one product, reconciled to the ledger.
04
Extend
Early default, prepayment, more products and the CECL input export.
05
Support
Monthly runs, definition changes under version control and new sources as products launch.
Facts and sources
Accounting rules behind the CECL data set
Short facts on expected credit loss accounting that shape the loan-level inputs, each linked to its primary source.
The interagency policy statement on allowances for credit losses, issued by the Federal Reserve, FDIC, NCUA and OCC, describes how expected credit losses are measured under the CECL methodology in FASB ASC Topic 326.
Source: Federal Reserve SR 20-12, revised April 21, 2023FASB's ASU 2022-02, Troubled Debt Restructurings and Vintage Disclosures, eliminated the troubled debt restructuring recognition and measurement guidance for creditors that have adopted ASC Topic 326.
Source: Federal Reserve SR 20-12, revised April 21, 2023In the Federal Reserve's capital rules, adjusted allowances for credit losses are the allowances that reduce the amortized cost basis of financial assets to the amounts expected to be collected, as determined under GAAP.
Source: 12 CFR 217.2 (eCFR)
Voice of success
We don't just build products; we forge lasting partnerships. See how we've helped industry leaders transform their vision into technical reality.
“I can clearly see how Agnotic has a unique way of handling end-to-end development. They are always active on quick chat and provide support quickly.”
Aaron Phelan
Founder, Benchmark
“Agnotic combines deep technical expertise with strong domain knowledge. They understand the business context, anticipate challenges, and make collaboration smooth and effective.”
John Pasmore
Founder, Latimer
“Agnotic is the best technical team we evaluated. Their engineering excellence made our work dramatically easier and allowed us to stay focused on what matters most for maternal care outcomes. They took full ownership of the technical execution, and we are always happy to continue working together.”
Kim Smith
Founder, My Lauren
Related pages
Dashboards, models and governance
This page covers the loan-level data and methods. These pages cover what is built on top of them.
BFSI analytics dashboard software
Executive dashboards across banking and insurance data, built on layers like this one.
Credit risk scoring software
Scoring models with their governance, backtesting and fair lending monitoring.
Loan origination software
The origination system that sets each loan's cohort and attributes.
Risk and compliance reporting software
GRC and regulatory reporting that consumes governed lending data.
Dashboard development
Custom dashboards and BI builds for any team.
More in Financial Services
More Financial Services Solutions
Explore other production-grade engineering platforms we deliver across financial services.
Frequently Asked Questions
Lending analytics cost depends on how many source systems hold loan history, how far back that history goes, how clean migrated loans are, the number of products, and whether the CECL input export is in scope. We size it on the scoping call after seeing your data.
It depends mostly on data history and quality. We agree a fixed first release on the scoping call, typically monthly snapshots and vintage views for one product.
Where the servicing system keeps payment history, we rebuild it by replaying payments and events against each loan's schedule to reconstruct month-end status. Where history is gone, we start snapshots now and mark the gap clearly.
No. Your finance team or model provider owns the expected credit loss model. We build the governed, reconciled loan-level input set it consumes, so every input traces to a loan and a source system.
Model validation, drift monitoring and fair lending testing are covered on our credit risk scoring software page. Lending analytics supplies the loan-level data those reviews use, with definitions and lineage they can rely on.
Yes. Lending analytics for startups starts with the snapshot table from the first loan, so vintage and early default views already exist when a funding partner or investor asks for them.
We build on your warehouse and BI tools where you have them. The logic lives in version-controlled SQL models, so the metric definitions do not depend on any single dashboard tool.
Industries