Reading the prepayment section

View as Markdown Open in Claude Open in ChatGPT

The Prepayment section shows how quickly your cohort (the loans your segment selects) returns principal earlier than its schedule requires, at each loan age. That early principal comes from two behaviors, loans paid off in full before maturity and partial prepayments on loans that stay open, and Vintage counts both as one measure: unscheduled principal returned.

Speeds are stated the standard way. The single monthly mortality (SMM) is the share of the balance that prepays in one month. The conditional prepayment rate (CPR) is the same speed annualized. The chart and its figures are explained below; the measurement is defined in full in Prepayment speeds.

The speed chart

The section has one chart: the cohort’s prepayment speed by loan age. A control above it switches between CPR by age (the default) and SMM by age. The horizontal axis is Loan age (months); the vertical axis is CPR (annualized) or SMM (monthly).

The line is solid where your own data backs it and dashed where it is projected to the end of the cohort’s term. The plotted line is smoothed for readability: each age’s speed is a trailing three-month average, weighted by the loans measurable at each of those ages, and a short gap of up to three months between well-backed ages is bridged on a straight line and drawn dashed. The projection eases from the last reliably measured speed toward the cohort’s long-run speed (the loan-weighted average over the most seasoned third of the ages it observed) and holds there to term, so prepayment does not stop at the last age your history reaches.
The bars behind the line are Loans measurable for prepayment at this age: how many loans could actually be measured for a speed there. That is often fewer than the loans on the book at that age, because a loan needs a reported schedule to be measured (see below). The point where the solid line ends is set by this measured count too: the line is drawn as observed through the last age at which at least 20 loans could be measured, and projected past it, even where a few loans still report. On a book where few loans report a schedule, the line therefore ends earlier and the bars stand lower.
Every mark on the chart is a measured or projected rate, so nothing on it rests on an assumed payment schedule.

Hover or tap a point for its speed and its measured loan count. As on the loss chart, no sentence describes a single point. A cohort that prepaid nothing at all is drawn on a full-scale axis, 0 to 100%, so a zero speed reads as zero. Otherwise the axis follows the speeds with modest headroom and never runs past 100%.

The headline speed

Above the chart, Avg annualized prepayment gives the cohort’s speed as one figure, captioned CPR (annualized). The same figure appears in the screen’s header as Avg Annualized Prepayment.

It is one annualization of the cohort’s pooled monthly speed, not an average of the per-age CPRs:

\text{SMM}_{\text{pooled}} = \frac{\sum \text{unscheduled principal}}{\sum \left(\text{starting balance} - \text{scheduled principal}\right)} \qquad \text{CPR} = 1 - \left(1 - \text{SMM}_{\text{pooled}}\right)^{12}

Both sums run over every loan-month in the cohort that could be measured, using the raw per-age figures rather than the smoothed line on the chart. The denominator is each month’s starting balance less the principal the schedule called for that month: the balance that was free to prepay. In words: of all the balance that could have prepaid, this share did each month, compounded over a year. A pooled SMM of 1% is a CPR of about 11.4%, not 12%.

When the speed rests on very little (fewer than 10 loans were measured at the last age the chart draws as observed), the caption reads CPR (annualized). Thin history, directional only.

What a speed needs: a schedule basis

Each month’s speed is the unscheduled principal returned that month divided by the balance at the start of the month less that month’s scheduled principal. The denominator needs a schedule basis: a scheduled principal figure you report (Scheduled Principal Due or Scheduled Principal Paid, on a snapshot or a transaction ledger), or a Scheduled Payment Amount together with an Actual Payment Amount in the same month. Vintage never reconstructs a schedule from balances, an interest rate, or a payment amount, so a loan with no schedule basis is left out of the speed measure entirely, even if it reports a prepayment amount directly. It still counts toward credit loss, which needs no schedule. Where the unscheduled principal comes from, and every exclusion rule, is in Prepayment speeds.

When the speed reads as a dash

A speed that cannot be measured shows a dash, never a confident 0.00% that would make a cohort look like it never prepays. There are exactly two reasons, and the screen names the one that applies, because they call for different fixes:

Reason	What the section’s headline says	What the chart area says	What fixes it
No loan in the cohort reports a schedule basis.	No prepayment basis: add scheduled-principal or payment columns	No per-age prepayment speeds for this cohort yet. Speeds need loans that report a scheduled-principal basis.	Map a Scheduled Principal Due or Paid column, or a Scheduled Payment Amount with an Actual Payment Amount.
The loans report a schedule, but the cohort has been seen in only one month.	Needs two consecutive months of data	No per-age prepayment speeds for this cohort yet. Speeds are measured from one monthly snapshot to the next, so they appear once this cohort has two consecutive months of data.	Upload a snapshot for the following month.

The header figure’s caption says the same thing more briefly: No scheduled principal to measure against, or Needs two consecutive months of data.

The second reason is the ordinary state after a first upload of a single month. It needs a consecutive month: for a book reported quarterly, the next quarter’s snapshot does not resolve it.

Both reasons describe the cohort on screen, not your uploads as a whole. A column you mapped for most of your book can still be absent from the particular segment you are looking at.

When prepayment cannot be measured, the rest of the screen runs its projections at zero prepayment and says so beside each figure it affects: under the loss curve, in Pricing, and above the month-by-month projection.

Loans whose speeds reflect full payoffs only

A loan with a schedule basis but no way to see partial prepayment (no partial or unscheduled amount, no actual principal, no payment totals) still belongs in the speed measure. Every loan is watched for a full payoff, so a loan that never paid off is a measured zero, not missing data. What cannot be seen for that loan is its partial prepayment, so its contribution to the speeds reflects full payoffs only. Adding a Partial Prepayment Amount or Unscheduled Principal Amount, or Actual Principal Paid, alongside the scheduled principal you already report, unlocks partial measurement.

Payoffs Vintage infers

Many core systems drop a loan from their export once it closes, without reporting a payoff. So when a loan stops appearing in your snapshots before your portfolio’s latest snapshot while it still owed a balance and had not reached maturity (with a three-month buffer before maturity), Vintage treats it as paid off in full, in the month after it was last seen, at its last reported balance. Only loans your data leaves open are inferred this way; the full rule is in Payoffs and exits.

Months the measure skips

A rate needs a starting balance Vintage actually observed. So three kinds of month are left out of the speed measure: months inside a gap in a loan’s reporting, the first month after such a gap, and the first month Vintage sees a loan that entered your data already seasoned. A prepayment amount reported on one of those months is set aside rather than spread or guessed. See Missing and unreported data.

What this section does not show

The section measures a speed and nothing else. It shows no remaining-balance curve and no cohort expected life: those would be projections of one representative loan on an assumed schedule, sitting among measured figures. The expected life of the loan you are pricing appears in Pricing as its Adjusted term. See Reading pricing results.

Prepayment here is the cohort’s historical speed. There is no refinance model that varies prepayment with where rates move. See What Vintage does not do.