The tape already knows

Your MCA book can look healthy while every new vintage is getting worse.

A monthly origination cohort, built from the loan tape, is how you stop managing an average and start watching a class of deals from the day cash goes out to the day it comes back, or does not.

Monday morning. The dashboard is polite. Collection rate: 91%. Delinquency: “in range.” The warehouse is still in-formula. Nobody is panicking.

Then someone slices the loan tape by funding month.

January’s class, at month four, had cumulative losses around 3%. June’s class, at the same age, is already at 5%. The blended book never showed it. New deals always look clean: they have not had time to break. Old deals look slow because most of their cash already arrived. The average was a compliment the book did not earn.

That gap is why credit-card issuers and auto lenders have used vintage analysis for decades, and why serious MCA shops are catching up. You do not manage a mixed pile of advances. You follow a monthly origination cohort (every deal funded in the same month) as it ages.

How the report walks

The tape is the ingredient list. The schedule is the recipe. The curves are the taste test. Unit economics is the decision.

01   Start with the class, not the pile

What a monthly origination cohort actually is

A merchant cash advance is a lump of cash to a small business today, paid back from future receipts. Fund $10,000 at a 1.35 factor and they owe $13,500. Origination is the day the money leaves. A cohort is every advance that left in the same month. Think of a graduating class.

Industry practice is to lock that class and watch it at the same age: months on book, or MOB. January at MOB 4 is comparable to February at MOB 4. That like-for-like view strips out the distortion of mixing deals originated under different standards, in different months, at different stages of life.

How MCA vintages typically season

Month 1   losses show up as a small number

Months 3–6   losses climb

After that   flatten. Most deals have paid off or already defaulted.

If half of a cohort’s eventual losses typically appear by month 4, a $300,000 loss at month 4 is a signal that ultimate losses may land closer to $600,000, while there is still time to tighten underwriting, reprice, or cut a channel.

02   The tape has to be able to talk

The loan-tape schema a cohort actually needs

The loan tape is the file of every advance: who got money, when, how much, what they owe, what they paid, whether they are late, and whether the funder stopped expecting the rest. Schema is just the columns. In practice this is often two files joined by an advance ID: a static origination table, and a payment or monthly snapshot table.

Volume rule: fewer than about 10–15 deals in a month? Use a quarterly vintage. One default can swing the collection rate by 10 points.

03   One origination. Then a calendar.

Why a cohort schedule beats a blended snapshot

Once the tape can name the class, you build a schedule. Rows are months on book. Columns are what happened to that same January group: cash collected, income recognized, balances still out, deals that went late, deals written off. You originated once. Everything after that is performance.

Keep the denominator honest: divide by original funded amount, not whatever is still outstanding this week. Hard defaults lag. Slow pay (missed remits, partial ACH) shows up at month two. By month six the only tools left are collections and write-offs.

Here is that like-for-like view. January and June, same months on book. January stays on a healthy collection path. June does not. The blended 91% never showed this.

Cumulative collections as a % of contracted payback. Teaching example.

04   Five shapes. One class.

The performance curves you extract from the schedule

Plot the schedule and you are not inventing new data. You are watching the same January class do five different things as it ages. Healthy MCA vintages typically see collection rates rise and then plateau near 90–95% of contracted payback. A curve that flattens early, or a new vintage that sits below last year’s at the same MOB, is the book talking.

Same January class on a calendar. Collections come in early. Write-offs stay a thin monthly line.

Teaching example. Thousands of dollars by months on book.

05   The point of the report

From those curves to unit economics: per loan, over its life

This is why the report exists. New volume can hide later losses. A life view of one vintage cannot. For each $1 originated: what cash came back, what was lost, what it cost to originate and service, and what was left as profit over the whole life of the deal.

Timing is not a footnote. A 1.35 factor collected in six weeks is a different business than a 1.35 factor collected in ten months. Factor rate alone is not profitability.

Write-offs are already inside collections. Cost lines are example operating assumptions.

Back to Monday morning. The 91% collection rate was not a lie. It was an average. January had already done its job. June had not. The loan tape had both stories the whole time. Cohort analysis is simply the decision to read them separately, and to let the curves, not the headline, tell you whether the next dollar you originate is worth originating.

Keep reading

MCA factor rate & annualized yield
Why the same 1.35 factor is a different return at 42 days than at 10 months.

MCA pricing & capital requirements
How expected loss, warehouse cost, and equity become a factor.

MCA borrowing base: eligibility & availability
Why delinquent and charged-off paper never sees the advance rate.

Sources & further reading

Experian, Vintage Analysis 101  ·  Vector, MCA borrowing base: eligibility & availability  ·  Vector, MCA pricing & capital requirements  ·  Vector, MCA factor rate & annualized yield