What is Claude CRE refinance sizing maximum loan amount analysis? It is the workflow of using Claude to determine the largest loan amount a lender will fund on a stabilized commercial real estate property by running three constraint tests in parallel: debt service coverage ratio (DSCR), debt yield, and loan-to-value (LTV), then identifying which constraint binds and structuring the proceeds around it. Refinance sizing is one of the highest-leverage uses of Claude for CRE operators because the difference between the binding constraint and the non-binding ones often equals seven figures of cash-out proceeds. For broader context, see our pillar guide on AI deal analysis scoring.
Key Takeaways
- Every CRE refinance has three sizing constraints (DSCR, debt yield, LTV) and exactly one binding constraint that determines the maximum loan amount.
- Claude correctly identifies the binding constraint in roughly 95 percent of stabilized refinance scenarios when prompted with all three lender hurdle rates.
- The seven-step workflow covers T12 normalization, lender constraint setup, parallel sizing tests, binding constraint identification, cash-out calculation, sensitivity analysis, and lender package generation.
- The most common operator mistake is sizing to LTV when debt yield actually binds, which leaves loan proceeds on the table or, worse, results in lender re-trade after rate lock.
- Always verify the lender's actual underwriting using the debt yield they cite, not the headline DSCR, since debt yield has overtaken DSCR as the binding constraint in 2026 for most stabilized assets.
Why Refinance Sizing Is Different From Workout Analysis
Sizing a new refinance is fundamentally different from analyzing a loan modification. In a workout (covered in our guide on Claude for CRE loan modification and workout analysis), you start with an existing loan and negotiate terms with the existing lender. In a refinance, you are sourcing a new loan against current property cash flow and current market values, then matching the largest possible loan to the lender's three sizing constraints. The math is forward-looking, not backward-looking, and the binding constraint can flip between deals on the same property type at different points in the cycle.
According to Mortgage Bankers Association data, debt yield has displaced LTV as the binding constraint on roughly 70 percent of stabilized commercial property refinances in 2026, up from 45 percent in 2023. The reason: floating-rate base rates remain elevated, lenders have tightened debt yield minimums to 9.0 to 9.5 percent on multifamily and 9.5 to 10.5 percent on industrial, and DSCR cushions have compressed.
Step 1: T12 Normalization
Before sizing the loan, Claude needs a clean trailing twelve-month (T12) operating statement. Most operator T12s have non-recurring items (insurance true-ups, deferred maintenance one-timers, vacancy from a single tenant rolling) that need to be normalized for lender underwriting. Lenders calculate underwritten NOI by stripping these and applying market-level vacancy and management fees.
Prompt Claude with: "Review the attached T12 and produce a normalized NOI calculation. Strip non-recurring items including [list specific items]. Apply market-level vacancy of [X] percent (lender standard for this property type), management fee of [Y] percent of effective gross income (lender standard), and replacement reserves of [Z] cents per square foot per year. Output the normalized NOI and a reconciliation table showing each adjustment."
Step 2: Set Up the Three Lender Constraints
Get the actual term sheet from your refinance lender. The constraints will be: minimum DSCR (typically 1.20x to 1.30x on multifamily, 1.30x to 1.40x on industrial, 1.35x to 1.45x on retail), minimum debt yield (typically 9.0 to 9.5 percent multifamily, 9.5 to 10.5 percent industrial, 10.0 to 11.0 percent retail), and maximum LTV (typically 65 to 75 percent for stabilized assets in 2026).
Also confirm the underwriting interest rate (most lenders use a stress test rate that is 25 to 50 basis points above the actual rate), the amortization period (25 or 30 years for most product types), and any interest-only provisions that affect the DSCR calculation.
Step 3: Run Parallel Sizing Tests
This is the prompt that earns the workflow its keep. Claude runs all three sizing tests simultaneously and outputs the maximum loan amount each constraint allows, then identifies which is binding.
Prompt template: "Calculate the maximum loan amount under three sizing tests: (1) DSCR-constrained: NOI of $[X] divided by minimum DSCR of [Y]x equals maximum annual debt service. Convert to maximum loan using underwriting rate of [Z] percent and amortization of [N] years. (2) Debt yield-constrained: NOI of $[X] divided by minimum debt yield of [W] percent equals maximum loan amount. (3) LTV-constrained: appraised value of $[V] times maximum LTV of [U] percent equals maximum loan amount. Output a table showing each constraint, the loan amount each allows, and identify which constraint binds (the smallest of the three)."
Step 4: Identify the Binding Constraint
The binding constraint is the smallest of the three loan amounts from Step 3. Claude will identify it explicitly in the table, but the operator decision starts here. If debt yield binds, the only path to higher proceeds is increasing NOI (rent growth, expense reduction, lease-up). If LTV binds, the path is a higher appraised value (cap rate compression, capital improvements). If DSCR binds (less common in 2026), the path is a lower interest rate or longer amortization.
Claude's value: it makes the diagnosis explicit and quantifies the gap between the binding and non-binding constraints. If LTV allows $24M but debt yield allows $21M, Claude tells you exactly why $3M of proceeds is being left behind and what NOI growth is required to close the gap.
Step 5: Calculate Cash-Out Proceeds
Once the maximum loan is set, calculate cash-out: max loan minus payoff of existing senior debt minus closing costs (typically 1.5 to 2.5 percent of new loan) minus any reserve requirements (lender-mandated tax, insurance, or capital reserves at closing).
Prompt Claude with: "Calculate cash-out proceeds. New loan amount: $[max from step 4]. Existing senior debt payoff: $[X] (include defeasance or prepayment penalty if applicable). Closing costs: 2 percent of new loan. Reserve funding: $[Y] for tax/insurance escrow plus $[Z] for capital reserves. Output: gross loan proceeds, total deductions itemized, net cash-out to borrower, and the implied cash-on-cash return on remaining equity using year 1 projected cash flow."
Step 6: Run Stress Sensitivities
Lenders almost always re-trade between application and closing if the property's NOI shifts. Stress test the sizing against three scenarios: NOI drops 5 percent (cap rate moves modestly, vacancy ticks up), NOI drops 10 percent (a single tenant rolls or rent collections soften), and the underwriting rate moves up 50 basis points (rate volatility through closing window).
For each, recalculate the binding constraint and the max loan. The output: a sensitivity matrix showing how loan size moves with NOI and rate. This becomes the negotiation tool with the lender if you want to lock in a rate buy-down or a lower DSCR threshold.
Step 7: Generate the Lender Package
The last step converts the model into the document the lender wants: a clean executive summary, the normalized T12, the sizing memo showing the binding constraint, the cash-out calculation, and a clear ask. Most operators waste 8 to 12 hours formatting this. Claude does it in 20 minutes.
For complex capital stacks involving subordinate debt, see our guide on Claude for preferred equity and mezzanine debt analysis.
The AI Consulting Network helps CRE operators build refinance sizing playbooks where the same Claude workflow produces consistent, lender-ready output across every property in the portfolio.
What Claude Does Not Replace
Claude does not replace lender relationships. The actual rate, the willingness to flex on a covenant, the speed of execution, these all come from the relationship the operator has with the credit team at the lending institution. Claude tells you what loan amount the property supports under the lender's stated constraints. The negotiation that gets the lender to actually fund that amount is still human work.
Frequently Asked Questions
Q: Why is debt yield more important than DSCR in 2026?
A: Debt yield (NOI divided by loan amount) is interest-rate-independent, while DSCR depends on both the rate and the amortization period. As rates rose 2022 to 2024, lenders moved to debt yield as the primary constraint because it gave them a measure of leverage that did not move with their own pricing. By 2026, most stabilized refinances size to debt yield first.
Q: Can Claude size a refinance for a value-add property without stabilized NOI?
A: With caveats. The lender will want a stabilized pro forma NOI plus a re-trade-out (an interest reserve to cover debt service during lease-up). Claude can model this as a bridge loan plus permanent take-out, but a value-add bridge loan requires different prompts than a stabilized refinance. The seven-step workflow above is for stabilized properties.
Q: How accurate is Claude's max loan calculation versus a banker's actual quote?
A: Within 2 to 4 percent typically. Most variance comes from the lender applying a slightly different stress rate or stricter vacancy assumption than the prompted one. Use Claude's output as the negotiation anchor, not the final number.
Q: Should I use ChatGPT or Claude for refinance sizing?
A: Either works for the math, but Claude's longer context window matters when the term sheet, T12, rent roll, and existing loan documents all need to be in the same conversation. Claude Opus 4.7 (with its 1 million token context window) holds the full document set without truncation. For a fuller comparison, see ChatGPT vs Claude debt analysis.
Q: What happens if the binding constraint changes between application and closing?
A: The lender re-sizes. If LTV was binding at application but debt yield binds at closing because NOI softened, your max loan drops to whatever debt yield allows. This is why Step 6 (stress sensitivity) matters: knowing the gap between constraints lets you anticipate re-trades before the lender raises them.