The 5.85%-Cap-That's-Really-4.95% Pattern
A short methodology note on the systematic gap between broker pro forma cap rates and recast cap rates on multifamily deals — and what to do about it.
We ran 100 multifamily deals through structured underwriting in Q1. The median gap between broker asking cap and our recast cap was 80 basis points. The top decile gap was 280 bps.
The mechanism is consistent across deals. Brokers don't fabricate numbers — they make assumptions:
- Vacancy modeled at "stabilized" 5% regardless of trailing actuals
- Rents stepped to "market" on every unit
- Management fee stripped out (you're sophisticated, you'll self-manage)
- Replacement reserves at zero or $150/door
- Property taxes at pre-sale basis (no reassessment trigger)
Each assumption, individually, is defensible. Stacked together, they understate true year-one expenses by 12-18% on the median deal.
The fix is process discipline: always recast to TTM actuals with full institutional expense load (3.5% management fee, $300/door reserves, post-sale tax basis), and always present both numbers side-by-side in the IC memo. The gap between them is itself the most useful metric.
Of the 100 deals we ran, the six we pursued were not the lowest-priced — they were the ones where the broker cap and the recast cap were closest. Cheap was cheap for a reason 70% of the time.
The full methodology with the deal-by-deal breakdown is in Lessons from Analyzing 100 Multifamily Listings.
Share kit
Post this to LinkedIn
Tip: copy the post text first, then click "Open LinkedIn share dialog" — paste into the dialog that opens. LinkedIn auto-renders the URL preview.
More Field Notes
Phoenix Multifamily — The Supply Story Is Decelerating
Phoenix multifamily deliveries peaked at 24K units in 2024. 2026 projects ~13K. The supply pressure that broke rents in 2024-2025 is materially abating — but the price discovery hasn't caught up yet.
Why We Build Tools That Don't Replace Judgment
A short note on the design philosophy behind LargeKite — AI augments analyst capacity rather than replacing committee judgment, and why that distinction matters more than the productivity gain.
OM Extraction: 47 Seconds, Four Red Flags
Walking through a real OM extraction on a 48-unit Tampa multifamily — the structured output, the red flags surfaced automatically, and where the extraction agent saved an analyst about three hours of work.