CREDITVINTAGE / RESEARCH CONTROL PANEL

Know what was known.

SYNTHETIC EXAMPLE

Application-time features, 180-day outcome maturity and chronological train / calibration / test windows. The default demo uses generated loans and makes no claim about a real lender.

Out-of-time loans120 2024 Jul–Dec vintage

Observed 90+ DPD27.5% six-report horizon

Calibrated Brier0.196 constant baseline 0.202

Review capture21.2% at 10% fixed capacity

Vintage drift

Observed default rate and predicted mean by origination month.

2024-07

30.0%mean PD 13.8% · n=20

2024-08

25.0%mean PD 14.9% · n=20

2024-09

35.0%mean PD 17.8% · n=20

2024-10

40.0%mean PD 12.9% · n=20

2024-11

15.0%mean PD 15.2% · n=20

2024-12

20.0%mean PD 13.1% · n=20

How to read this

ROC-AUC 0.655; average precision 0.465. The vintage-block bootstrap 95% ROC-AUC interval is 0.522–0.771; only six test months are available.

Raw logistic Brier 0.181; calibration increased it to 0.196 in this run. Calibration is reported as a measured step, not assumed to improve every sample.

Scenario expected loss $132,386.46 assumes an unchanged LGD of 45% and origination principal as exposure. It is not a measured recovery loss, accounting provision, or regulatory capital estimate.

Training outcomes are mature before calibration applications; calibration outcomes are mature before test applications. Test labels never enter fitting or threshold selection.

Risk buckets

Bucket 1 has the highest predicted probability. Rates are descriptive and may be noisy.

Bucket	Loans	Mean PD	90+ DPD	Scenario EL
1	12	36.4%	58.3%	$36,095.07
2	12	22.2%	33.3%	$19,780.87
3	12	18.5%	25.0%	$14,854.05
4	12	18.5%	25.0%	$17,484.36
5	12	15.1%	33.3%	$13,882.17
6	12	13.3%	33.3%	$12,419.65
7	12	13.3%	25.0%	$11,854.96
8	12	4.4%	8.3%	$2,685.44
9	12	3.6%	25.0%	$2,713.40
10	12	0.9%	8.3%	$616.47