Auto ABS credit, one measure at a time

Short narrated videos. Each one follows a handful of example loans through the arithmetic, shows the usual mistakes, then runs the same calculation on our loan-level data. They run in order: each uses only ideas from the ones before it.

20 of 20 published, each with subtitles and a written page

Start Here  No. 1 · 2:22

Static pools, vintages and months on book

Frame from the static pools, vintages and months on book explainer

Why loans are grouped by the month they were made and compared at the same age.

The series

In learning-path order. Each page has the video, the worked tables, the common pitfalls and where to find the measure in the dashboard and decks.

  1. 1

    Static pools, vintages and months on book

    Why loans are grouped by the month they were made and compared at the same age.

    Video, 2:22

  2. 2

    Delinquency buckets and DQ rates

    Days past due, the 30+ and 60+ buckets, and why a rate weighted by balance differs from one by count.

    Video, 2:26

  3. 3

    Censoring

    Why young vintages stop early on a chart and are left blank, not shown as zero.

    Video, 2:09

  4. 4

    Charge-off frequency, severity and cumulative net loss

    How charge-offs and recoveries net into cumulative net loss, and how it splits into how often loans default and how much each one loses.

    Video, 3:03

  5. 5

    Loss to liquidation

    Net loss over all the principal that has left a group of loans, and why paydown from before a loan entered its trust still counts.

    Video, 1:48

  6. 6

    Ghost curves and vintage ranks

    Reading this year's vintage against faded past vintages, and its rank at month 12 and month 24.

    Video, 1:50

  7. 7

    Roll rates and transition matrices

    Ten loans move between payment statuses for a month, and the dollars are counted into a from-to table.

    Video, 2:01

  8. 8

    Following a dollar: the Markov flow

    Chaining monthly roll rates into the path from a missed payment to a charge-off.

    Video, 2:05

  9. 9

    σ above baseline in the transition heatmap

    How the transition heatmap marks an unusual month against the 2017-2019 control period.

    Video, 2:05

  10. 10

    Delinquency decomposition

    Splitting a delinquency rate into how many loans enter a bucket and how long they stay there.

    Video, 1:52

  11. 11

    Year-over-year overlays and seasonality

    Comparing the same month across years, and why delinquency follows the calendar.

    Video, 1:40

  12. 12

    FICO corridors

    Early delinquency by FICO band, read against each band's own history.

    Video, 2:45

  13. 13

    Issuer against its own history

    Where an issuer's latest number sits within the range of its own past.

    Video, 2:35

  14. 14

    Collateral: LTV and vehicle age

    Loan-to-value and vehicle age, and why both drive how much a charged-off loan loses.

    Video, 2:39

  15. 15

    Payment extensions

    The extension rate, the payments moved, extensions by starting delinquency, and the share of each vintage ever extended.

    Video, 2:12

  16. 16

    Extension Default Rate

    The share of extended loans that are 30 or more days late, or charged off, two months after their moved payments end.

    Video, 2:08

  17. 17

    Extended vs untreated

    Comparing extended loans with similar loans that were not extended, and what that comparison can and cannot prove.

    Video, 1:46

  18. 18

    Bridges and waterfalls

    Splitting a month-to-month change into the parts that made it.

    Video, 2:02

  19. 19

    Prepayment and pool paydown

    Scheduled and actual principal, the monthly and annual prepayment rates, and the pool factor.

    Video, 2:33

  20. 20

    Bayesian nowcast curves

    Estimating this month's delinquency before every issuer has filed, and how the estimate moves as filings arrive.

    Video, 2:31

Every real-data example comes from SEC Form ABS-EE loan-level filings.