Private credit at the late-cycle turn

Late in a credit cycle, the most useful signal is rarely price. Spreads compress slowly and recover suddenly, which means the market clears at levels that say little about how individual loans were underwritten. The divergence worth tracking is between what managers are paid to take and what their documents actually let them control.

Where discipline erodes first

Underwriting standards rarely fail loudly. They erode in increments that are individually defensible and collectively decisive. In manager diligence, four patterns deserve disproportionate attention:

• Covenant drift — maintenance covenants replaced by incurrence-only structures, then incurrence baskets widened deal over deal.
• Adjusted-EBITDA inflation — addbacks that began as one-time items recurring across consecutive amendment cycles.
• Amend-and-extend as a default — extensions that defer recognition rather than restructure economics.
• PIK toggles exercised quietly — cash-pay conversion that changes the risk profile without changing the mark.

What the gap implies for allocators

None of these patterns is disqualifying on its own. The question is whether the manager can show you, loan by loan, where they sit on each — and whether their marks move when the documents move. A book that re-marks only on payment default is reporting history, not risk.

The price of a loan tells you what the market will pay. The documents tell you what you actually own. Late in the cycle, the second number is the one that pays for diligence.

Allocator workplans we build for private credit reviews weight document-level sampling over portfolio-level statistics for exactly this reason: the portfolio view is the manager’s narrative; the sample is the evidence.