ArcFi
LENDING MARKETPLACE
THE ARCFI UNDERWRITING MODEL

Your business data. One score.
A rate you actually earned.

ArcScore analyzes four real-time data sources to generate a risk-tier score. ArcScore sorts borrowers into risk tiers from safest to highest-risk, so investors can match opportunities to their risk appetite.

How ArcScore is built

Four data inputs. One risk number. Updated in real time every time you apply.

How your rate is set

ArcScore maps to one of four risk tiers. Your tier determines your APR — from 8% for Tier A to 18% for Tier D.

Fair lending by design

ArcScore excludes all prohibited basis variables. Adverse action notices are generated for every decline. Periodic disparate impact testing is built into operations.

DATA SOURCES

Four real-time inputs. One score.

SECTION 1

Bank Account Data (via Plaid)

LIVE DATA · UPDATED AT APPLICATION
  • 90-day average daily balance
  • Deposit regularity and source concentration
  • NSF and overdraft frequency
  • Debt service coverage ratio
  • Revenue trend direction (growing / stable / declining)
  • Expense composition and burn rate

SECTION 2

Receivables Quality (via Codat)

ACCOUNTING INTEGRATION · REAL-TIME
  • AR aging schedule (current / 30 / 60 / 90+ days)
  • Customer concentration (is one customer >50% of AR?)
  • Historical dilution rate (how much AR doesn't collect?)
  • Average days to collection by customer
  • Counterparty payment behavior patterns

SECTION 3

Commercial Credit History

BUREAU + PUBLIC RECORDS
  • Payment behavior with existing creditors
  • Public records and judgments
  • UCC lien search — existing encumbrances
  • Trade reference verification
  • Counterparty entity verification and standing

SECTION 4

Application Data

VERIFIED AGAINST ALL SOURCES
  • Business structure and time in operation
  • Industry classification and seasonality
  • Stated use of funds
  • Invoice authenticity cross-check
  • Customer counterparty verification

A NOTE ON MODEL MATURITY

ArcScore is a defensible methodology — not yet a validated model. It becomes genuine intellectual property only after 500–1,000 loans have completed full repayment cycles and produced statistically meaningful loss data. ArcFi does not overclaim model accuracy at launch. Every one of our first 100 loans includes mandatory manual underwriting review regardless of score output. We will publish vintage performance data as it accumulates.

FAQ

Questions people actually ask

All ArcFi loans are short-term receivables-backed advances, typically 45 to 75 days. The term matches the underlying invoice payment timeline.

Still have questions?

Join the waitlist and our team will reach out.