What It Evaluates — Proprietary AI Underwriting Language
Six Lenses. One Verdict.

Every grade traces back
to defined inputs.

The engine is a documented scoring system, not a black box. Each of the six metrics has named inputs, a known scale, and a defined path to improvement. If a deal scores low, you know exactly which lever to pull.

composite.grade
# six lenses, composite output
sponsor_strength A−
funding_probability 82
stabilization_risk 28 / low
refi_probability 74
capital_fit 14 matched
improvement_set 2 ranked

composite A−
The Six Metrics

An instrument panel for every deal.

Each metric reads one dimension of the file. Read together, they tell capital sources whether to fund — and if not, exactly what to fix.

METRIC 02

Funding Probability Index™

A 0–100 score weighted across sponsor, asset, structure, and market conditions. It answers the only question that matters at the top of the funnel: what are the odds this specific file gets a written term sheet from a real capital source in the next 30 days?

// Inputs
  • Sponsor Strength Rating
  • Asset/market signal
  • Structure efficiency (LTV, DSCR, reserves)
  • Capital market depth for this profile
  • Deal urgency & close-by date realism
// Output
82/100Index 0–100
// Scale
High
// Improve
Tighten leverage one tier, or move close-by date by 2–3 weeks to widen the bidding lender set.
METRIC 03

Stabilization Risk Score™

A forecast of how likely the asset reaches stable, lender-accepted cash flow on the projected timeline. Lenders use this to size reserves, set interest-only periods, and decide whether to do the deal at all on value-add and ground-up files.

// Inputs
  • Lease-up velocity vs. market comps
  • Cap-ex burn schedule & contingency
  • Rent roll concentration
  • Market supply pipeline & rent growth
  • Macro overlay: rates, refi depth
// Output
28 · lowIndex 0–100 (lower = safer)
// Scale
Low Risk
// Improve
Add 12–18 months of interest reserves, or shift to a phased lease-up plan with conservative pro-forma.
METRIC 04

Refinance Probability™

Models the takeout before you enter. The engine forecasts the odds and timing of refinancing into long-term debt at the projected exit. Lenders care because it's their exit too — a deal with a soft takeout is a deal they price defensively.

// Inputs
  • Projected stabilized NOI & cap rate
  • Forward rate curve & agency appetite
  • Asset class refi liquidity (multifamily > office)
  • Sponsor balance sheet at refi date
  • Loan-to-stabilized-value ceiling
// Output
74/100Index 0–100
// Scale
Likely
// Improve
Lengthen IO period, or model a sale-then-refi path if forward agency execution is tight.
METRIC 05

Capital Fit Signal

The lender directory, filtered. From the engine's understanding of who funds what right now, it returns a ranked list of capital sources whose box this deal fits inside — not who their marketing says they fund, but who they actually closed in the last 90 days.

// Inputs
  • Lender's recent close history (90 days)
  • Asset class & deal size box
  • Sponsor tier accepted
  • Geographic footprint & state license
  • Structure preferences (recourse, IO, mezz)
// Output
14 matchedRanked list
// Scale
Targeted
// Improve
Match precision rises sharply with a tighter close-by date and full sponsor docs.
METRIC 06

Deal Improvement Set

The ranked, specific changes that move a borderline file into fundable territory before submission. Not generic advice — the exact moves, in order, with the projected grade impact of each. This is what closes the gap between a 78 and an 86.

// Inputs
  • Detected weak signals from passes 02–04
  • Lender box overlap analysis
  • Cost-to-impact ratio of each fix
  • Sponsor's ability to execute the fix
  • Time-to-fix vs. close timeline
// Output
2 rankedRanked actions, with impact
// Scale
Actionable
// Improve
Each item lists the change, the cost (real or opportunity), and the expected grade impact.
Translate Your Next Deal

Put your file in the language lenders fund.

Submit a scenario and see it scored, structured, and packaged the way capital sources actually underwrite — in minutes.