← Field notes

Signal series · Part 4 of 5 · July 28, 2026 · 5 min read

Inflection: anatomy of a bust-out week.

For most of its life a cultivated account is calm. Then, near the end, it changes character sharply. Measured against its own prior months, that is a regime break, and it is the loudest moment in the whole lifecycle.

Part 4 of a five-part series on the signals a retrospective review reads. Earlier: Ramp. Next: Ring.

The first three signals are quiet. Provenance, coherence, and ramp all live in the slow middle of an account's life, and none of them announces itself. The inflection is the opposite. It is the week the account stops behaving like the account it has been and starts behaving like something taking everything it can before it disappears.

What a regime break means

The key word is own. We are not comparing the account to a population, or to a fraud template, or to a threshold. We are comparing the account to itself. For twenty, thirty, thirty-six months it has held a baseline: a typical balance, a typical utilization, a typical mix of purchases, a typical payment behavior. A regime break is when several of those move at once, in the same short window, and land far outside the range the account itself established. What is dramatic for a normally-quiet account might be routine for a naturally-volatile one, which is exactly why the baseline has to be the account's own.

The shape of the week

The break tends to show a recognizable cluster of moves, arriving together:

  • Utilization runs to the limit. An account that sat at a modest fraction of its line for years pushes to the ceiling and stays there. Not a normal busy month, but the balance pinned against the limit.
  • Cash and cash-equivalents spike. Spending shifts toward the most fungible, least reversible forms: cash advances, cash-like purchases, moves that convert a credit line into something that can walk out the door and not come back.
  • Payment behavior flips. An account that paid down reliably starts paying the minimum, or paying just enough to free headroom for one more draw right before the end.
  • It happens fast. The whole shift compresses into days or a couple of weeks, against a history measured in years.

Any one of these can happen to a real member having a hard month. The signature is the combination, arriving at once, against a long calm baseline, in an account that also lacked provenance, lacked coherence, and let its appetite outrun its income. The inflection is where the earlier signals are confirmed, not where the case begins.

The loudest moment is the latest one

Here is the uncomfortable part. The inflection is the easiest signal to see and the least useful to see first, because by the time it arrives the decision that mattered has already been made. The credit was extended and the limit was raised months earlier, against an account that looked healthy the entire time. A monitoring approach that waits for the bust-out week is watching the moment the money leaves, not the moment it could have been kept. It is a receipt, not a warning.

That is the whole argument for reading the earlier, quieter signals. If the inflection is the first thing you notice, you have noticed a loss in progress. If provenance, coherence, and ramp were read across the account's history, the inflection is merely the confirmation you were already expecting, on an account that had been flagged well before the week it broke.

One account, then its neighbors

An inflection on a single account tells you about that account. But synthetics are rarely built one at a time, and a bust-out week rarely happens in complete isolation. The final note in the series is about what a confirmed inflection should make you do next: stop looking at the account, and start looking at the accounts around it.

Sources & notes

This is a conceptual, operator-facing description of one behavioral signal. It makes no factual claims about any specific institution, portfolio, or case, cites none, and describes no Delegate engagement or result. The mechanics it names (utilization, cash-advance behavior, minimum-payment shifts, change against an account's own baseline) are ordinary properties of account data. You can see the signal families run on a simulated book in the Portfolio Explorer.