# Origination and term Everything on the Modeling screen is arranged by **loan age**: how many months a loan has been on the book since it was made. A loan made in 2019 and a loan made in 2024 both contribute to "what happens at month six". To place a loan on that axis, Vintage needs two things about it: - **when it started**, so its age in any month is known, and - **how long it was written for**, its **term**, so its expected lifetime is known. A loan is **modeled**, meaning it enters the age-indexed curves, when it can be aged, has a term, has at least one snapshot row with a readable Snapshot Month and a Current Balance (a reported 0 counts), and was not already closed on the first row Vintage sees for it. The first two rules are set out below: where a loan's start and term come from, when Vintage estimates one, and when it refuses to. ## Age Age is counted in whole months from the loan's **origination anchor**, the month Vintage treats as the loan's start. The origination month itself is age 0. A loan originated in March 2023 is age 0 in March 2023, age 1 in April 2023, and age 12 in March 2024. Aging is measured against one **as-of** date for the whole portfolio, the latest snapshot month across all your data, so every slice is on the same clock. See [Vintage analysis](/concepts/vintage-analysis/). ## The origination anchor ### When you provide an Origination Date A supplied **Origination Date** is always the anchor. It can come from an Origination file or a column on your snapshots. When both provide one, the Origination file's value wins; the full rules for files that disagree are on [Uploading month after month](/uploading/uploading-month-after-month/). ### When Vintage estimates it If a loan has no Origination Date, Vintage can still estimate its start from its **first appearance**: the first snapshot month in which the loan is reported. It does this **only for a loan that first appears after your organization's earliest snapshot month**. For such a loan, first appearance is a real signal. The loan was not in any earlier month's data, so it most likely started when it first showed up. Vintage ages it from that month, uses the **first balance it sees** as its original balance, and marks it **Estimated origination**. Treating the loan as originating when it first appears is an approximation, which is why it is labeled rather than silent. ### Why a loan in your earliest month cannot be aged Your **earliest snapshot month** is the first month of data across all of your organization's accepted uploads. A loan already present in that month, with no Origination Date, is the case Vintage refuses to guess. Suppose your first snapshot is January 2020 and two loans appear in it, neither with an Origination Date. One was made in December 2019; the other in 2014. In your data they look identical: both are there in the first month. Estimating both as originating in January 2020 would make a six-year-old loan look new, and every loss or payoff it had would be booked at the wrong age. So such a loan **cannot be aged**. It is **excluded from the age-indexed curves** and counted, with a note that a small file would unlock it. **A two-column file unlocks these loans:** An Origination file with only two columns, **Loan ID** and **Origination Date**, gives every earliest-month loan an anchor. Add **Original Loan Amount** as a third column and those loans also enter the loss curve measured as a percent of original balance. To add one: 1. Export Loan ID and Origination Date for the loans on your book, from your core system's loan-level data. 2. Start a new upload and add the file under the **Origination Files** card. 3. On **Remove PII**, designate the column that holds the Loan ID. Its values are scrambled the same way as on your snapshots, so they match your existing loans; the column's heading does not have to match. 4. Map the date column to **Origination Date**, confirm its date format, and finalize. Vintage then re-ages those loans and they enter the curves. ### When your earliest month moves The earliest snapshot month belongs to your whole organization, so it can move. Uploading older history moves it earlier; deleting the files that carried your first months moves it later. Either way Vintage recomputes the whole portfolio's modeling, because a loan that could be estimated before may now be an earliest-month loan, or the reverse. A loan that only looked new because older months were missing is reclassified honestly once they arrive. See [Upload History](/uploading/upload-history/). ## The original balance A loan's original balance is the **Original Loan Amount** you supply. Without one, Vintage uses the first balance it sees. For a loan aged from an estimated origination, that first balance is close to what was lent: the loan was first seen when it started. For a loan that was already seasoned when your data begins, the first reported balance is what was left after years of paying down. Vintage does not treat it as an original amount. Such a loan is held out of the loss curve measured as a percent of original balance, and counted, while its losses still count in every rate measured against the outstanding balance. See [Credit loss measurement](/concepts/credit-loss-measurement/). ## The term A loan's **term** is the number of months it was written for. Vintage takes it from: - a supplied **Term (Months)**, or, failing that, - a supplied **Maturity Date**, as the calendar months from the origination anchor to maturity. A term of zero, or a maturity date at or before origination, is not a usable term. A term that comes from a maturity date is recorded as **derived**, so it is never presented as one you supplied. A loan originated in March 2020 with a Maturity Date in March 2025 has a derived term of 60 months. ### Why a loan with no term is excluded A loan with neither a usable term nor a usable maturity date is **excluded from the curves**. Without a term, the loan's lifetime cannot be defined, and its ordinary run-off would be miscounted as prepayment. Because of this, a term is part of the base every modeled output requires. The Review never calls a loan ready for credit loss, payoff or prepayment when it has no term, so its readiness counts match what the curves will actually include. See [Modeling methods](/reference/modeling-methods/). ### What the term does in a measurement In measuring your history, the term has exactly two roles, and both are gates: 1. **It decides whether a loan is modeled.** No term, no curves. 2. **It sets the maturity buffer for inferred payoffs.** A loan last seen three months or less before the end of its term, and never seen again, is treated as having matured on schedule, not as having prepaid. See [Payoffs and exits](/concepts/payoffs-and-exits/). The term is **never** used to build an amortization schedule for a measurement. No curve Vintage measures from your history amortizes anything. A Maturity Date is a source of the term and nothing more. The term does feed the **projections**. The balance-weighted term of the cohort (the loans your segment selects) is the horizon its curves are projected to, and the term you enter in Pricing is the term of the loan being priced. See [Loan pricing](/concepts/loan-pricing/). ## Where you see origination and term On the **Review**, before you finalize, Vintage shows where each origination field comes from: provided, naming its column, or **will be estimated**. Loans in your earliest month with no anchor are flagged as excluded from the age-indexed curves rather than estimated. The same panel shows where the term comes from: a provided Term (Months) column, **derived** from a Maturity Date, or **missing**, flagged in amber. Naming a column is not the same as every loan having a usable value in it. A mapped Term (Months) column can still leave out loans whose term reads `0`. So the Review also states, per reason, how many loans are held out of the curves and what would bring them in. Among those reasons: loans that cannot be aged, loans with no usable term, and loans whose dates do not fit the column's declared format. See [Reviewing before you finalize](/uploading/reviewing-before-you-finalize/). ## Related - [Vintage analysis](/concepts/vintage-analysis/) - [Missing and unreported data](/concepts/missing-data/) - [Credit loss measurement](/concepts/credit-loss-measurement/) - [Payoffs and exits](/concepts/payoffs-and-exits/) - [Reviewing before you finalize](/uploading/reviewing-before-you-finalize/) - [Standard fields](/reference/standard-fields/)