Score a new credit product from launch day

Day-Zero scores it before the first default. As loans mature, zypl.Foundation learns each one in context and takes over scoring on your sign-off, once your own outcomes show it ranks better.

Test NowTalk to Sales

zypl.Foundation · demo cabinet

⌘J

Your bookThe cardPolicyOne applicantThe whole bookDoes it holdValidation

Columns in use

the rest left out, with reasons

Weights with a reason

6 / 10

written before any outcome

Derived ratios

built from your columns

Compounding pairs

treated as compounding

Weights · and the reason for each

0.80Times 90+ days late

NOT

RUO

NOT

NOD

NRE

NOO

Reason codes · strongest first

NOT
Times 90+ days late
Delinquency roll-rate pathway from moderate arrears to severe default
0.80
RUO
Revolving credit utilisation
Compound risk of near-limit credit lines combined with early arrears indicates imminent default.
0.65
NOT
Times 60–89 days late
Early delinquency roll-rate pipeline
0.55
NOT
Times 30–59 days late
Compound risk of near-limit credit lines combined with early arrears indicates imminent default.
0.45
a
Age
-0.35

Columns you will not lend on

Agein use
Real-estate loans and linesin use

Left out at decision time

nothing that is unknown when you decide

Bulletin № 01 · From day zero to handoverOne product · three stagesFrom the claim ledger

Headline · the outcome gap

A new product has no defaults of its own.

A fitted scorecard needs matured loans, and a new product, market or segment has none yet. Until a vintage matures, decisions run on rules borrowed from another book, and the cost of a wrong rule shows only when the defaults arrive.

01 · Day zero
Day 0
a Day-Zero scorecard, before the first default
Built from the data dictionary and feature statistics, and reviewed by your validator before launch. Its ranking quality shows once your own outcomes arrive.
02 · As loans mature
+2.1
Gini points at 100 matured loans
Comparator: TabICLv2 with eight members. Ranking quality, not calibration; outcome-stratified contexts. Gini = 2·AUC − 1.
03 · Proven on your outcomes
0.651 → 0.699
served AUC, Day-Zero alone to Handover
A retrospective replay between Day-Zero and zypl.Foundation — not a live deployment, and not yet with Zehn. Always serving zypl.Foundation reached about 0.711: Handover is a governed, reversible policy, not an accuracy gain over the foundation model. Switches occurred at application index 2,862 to 11,182.

Every number has a caveat. Each one is on the page.

Test NowEvery figure with its source

§ How it works · from day zero to handover

From day zero to your own outcomes.

01 · Day zero
Day zero: a scorecard before any repayment.
No outcomes yet. Send the product's data dictionary and feature statistics: a scorecard comes back with points, grades and reason codes, and a written reason for every weight, for your validator to review before launch.
What Day-Zero delivers
Day-ZeroGive Me Some Credit
Times 90+ days late0.80
NumberOfTimes90DaysLate
Delinquency roll-rate pathway from moderate arrears to severe default
Revolving credit utilisation0.65
RevolvingUtilizationOfUnsecuredLines
Times 60–89 days late0.55
NumberOfTime60-89DaysPastDueNotWorse
Times 30–59 days late0.45
NumberOfTime30-59DaysPastDueNotWorse
6 more weights, each with its reason
02 · As loans mature
Then switch on zypl.Foundation.
Each loan that matures joins the model's context, with no training run per product. zypl.Foundation scores beside the Day-Zero card and your current process. It reads any labelled table the same way: loans, or another outcome you want to predict.
Gini points over the strongest open model of its class
100 matured loans
+2.1
100–500 loans · pre-registered
+1.3
Full data
+0.8
11 public credit tables · full context
+1.7
Comparator: TabICLv2 with eight members. Ranking quality, not calibration; outcome-stratified contexts. Gini = 2·AUC − 1. At full data the gain comes from the serving procedure, not from the specialised weights. Measured on lending books. On other tables, test it on yours.
Every figure with its source
ContextIllustrative
Matured loans, as they arrive
LN-04122026-01repaidin context
LN-04172026-01repaidin context
LN-04332026-02defaultedin context
LN-04382026-02repaidin context
LN-04512026-03repaidin context
LN-04692026-03repaidin context
No training run · each loan is read as context
03 · Proven on your outcomes
Handover, on your sign-off.
zypl.Foundation takes over scoring when your own held-out outcomes show it ranks better, and your risk owner signs. Until then it scores in shadow, and service moves back if the evidence turns.
HandoverIllustrative
Shadow period
Your current process
serving
zypl.Foundation
scoring in shadow
Held-out outcomes
sealed until review
Validator
pack reviewed
Risk owner
sign-off pending
Service moves on your sign-off

Day-ZeroGive Me Some Credit

Times 90+ days late0.80
NumberOfTimes90DaysLate
Delinquency roll-rate pathway from moderate arrears to severe default
Revolving credit utilisation0.65
RevolvingUtilizationOfUnsecuredLines
Times 60–89 days late0.55
NumberOfTime60-89DaysPastDueNotWorse
Times 30–59 days late0.45
NumberOfTime30-59DaysPastDueNotWorse

6 more weights, each with its reason

Now showing · Day-ZeroStage 1 / 3

4 questions · head to headLaunch day, learning, validation, deploymentzypl.Foundation

With

zypl.Foundation.

Today

Borrowed rules.

01Launch day: A Day-Zero scorecard: points, grades, reason codes.; Policy rules and judgement from another book.
02Matured loans: Each one learned in context, no training run.; Wait for a vintage, then a modelling project.
03Validation: A written reason for every weight, up front.; A model the validator meets after the fact.
04Deployment: Inside your environment, on server CPUs.; Wherever the scorecard was built.

01Launch day

WithA Day-Zero scorecard: points, grades, reason codes.

TodayPolicy rules and judgement from another book.

02Matured loans

WithEach one learned in context, no training run.

TodayWait for a vintage, then a modelling project.

03Validation

WithA written reason for every weight, up front.

TodayA model the validator meets after the fact.

04Deployment

WithInside your environment, on server CPUs.

TodayWherever the scorecard was built.

§ End of pageSee it on a public book, then on yoursDemo · public lending books

See it on a public book, then on yours.

The demo walks Day-Zero on public lending books, then zypl.Foundation on the example or a labelled CSV of your own. A Day-Zero Readout reads one of your products from its data dictionary, within 48 hours.

Test NowRequest a Day‑Zero Readout

Day-Zero Readout

01Top drivers
02Expected ranking band
03Data-readiness gaps

What comes backWithin 48 hours

Launch day · a scorecardEvery matured loan · in contextYour servers · your outcomesYour validator · your sign-off