Point-in-time snapshot
A point-in-time snapshot is a copy of data that preserves what it showed at a specified date. Lenders use month-end snapshots to analyse each loan as it stood then, rather than rely on today's record, which may include information unavailable at the time.
Example
A point-in-time snapshot preserves each loan’s balance, days past due and status at month end. For illustration, a loan is 5 days past due in January and 0 in February; using February’s status in a January roll-rate analysis would substitute a later status for the one recorded at January month end.
Building and keeping month-end loan snapshots
A monthly loan-book view has one row per loan for each month end. The row holds the balance, days past due and status recorded for that date.
Keep the original version of a published month end. If it is corrected later, retain the restated version beside it so a report can be reproduced as it was first shown.
Using point-in-time data in credit analysis
Scorecard monitoring can compare decisions with later outcomes. Vintage curves and roll rates use month-end statuses. Provision calculations use them too.
For model development, use the information available at the time being reconstructed. Today's record may include facts that were absent then, making a model or rule look better than it was.
Questions we get asked
What does a point-in-time table do in a data vault?
In a data vault, a point-in-time table is a business vault helper that selects the satellite rows valid at a given date. It brings together the rows that apply to that date.
Is month-end data the same as decision-time data?
No. A month-end snapshot records a loan's balance, days past due and status on that date. Decision-time records show what information was available when a past decision was made.
Can a published month-end report be reproduced after a correction?
Yes, if the original and restated versions are both kept. The original lets the report be reproduced as it was first shown.
A point-in-time snapshot keeps what was known at a set date, and the article on data warehouses in lending connects month-end snapshots with decision-time records.