Looked fully underwritten. 20% of the downside was directional.
The model was not wrong. The basis was incomplete.
A representative pattern, not a customer engagement — Capital Refinery has no client case studies to publish yet. The figures below are illustrative of what the engine surfaces, on real-shaped fixtures.
Full IC memo prepared. Historicals in place. Downside case included. Coverage looked safe enough to keep moving. The downside looked covered.
What changed the decision was not another round of formatting or model cleanup. It was exposing that one part of the downside was still directional — and carrying too much of the answer.
1. What looked fine.
Full IC memo prepared and the recommendation was already stabilizing. Model built from historicals plus forward projections. Downside case included and presented cleanly. Coverage held around the mid-2x range in the initial downside framing. The deal was moving toward IC with no obvious signal that it should stop.
2. What was actually wrong.
One operating segment, roughly 20% of EBITDA, was projected from KPI assumptions instead of full statement-backed support. That segment was treated as stable enough inside the downside case. The IC narrative inherited that confidence even though the basis was still partial.
That portion of the downside was directional, not grounded. It was treated like fact anyway.
3. What would have happened.
If that segment underperformed, downside coverage moved materially lower. Worst-case DSCR compressed toward roughly 2.1x instead of holding where the memo suggested. Pressure showed up earlier than the original timing implied. The refinancing path became more fragile because the wrong downside had been carrying the approval case.
4. What Capital Refinery surfaces.
The directional share was flagged immediately instead of staying hidden inside one downside number. The downside moved immediately once the directional share was tested. The break point moved forward and the timing pressure became visible. The output forced a cleaner decision — pause or restructure before IC instead of continuing on confidence alone.
This is not faster modeling. It is faster decision clarity under pressure.
5. What changed fast.
Initial pressure test: same day. Iteration with the team: 2–3 days. Decision-ready framing: inside one week. This was not weeks of model iteration. It was one decision made visible.
6. What this means on your deal.
If part of your downside is directional, you do not actually know your downside yet. This is not a full rebuild. It is one decision, made visible. You do not need permission to test it. Run one deal. You’ll see it immediately.
Start with the file no one is questioning.
If part of your downside is directional, you do not actually know your downside yet. Bring us one deal — we'll show you the directional share, the earlier break point, and the decision the pressure test forces.