πŸ“ŠRisk Scoring

LendFriend provides transparent risk assessment to lenders, who decide what risk they're comfortable with. We follow Prosper and LendingClub's model: disclose risk clearly, let the market decide.


Risk Grades

Every loan receives a grade from A (minimal risk) to HR (very high risk):

Grade
Risk
Description

A 🟒

Minimal

Excellent history + strong ties

B 🟒

Low

Good history OR strong ties

C 🟑

Moderate

Some history, moderate ties

D 🟑

Elevated

Limited history or weak ties

E πŸ”΄

High

No history + large loan

HR πŸ”΄

Very High

High risk situation or recovery

High risk grades (E-HR) may not fund successfully. Borrowers should start smaller or build their network first.


How Grades Are Calculated

Grades are based on four factors:

1. Repayment History

  • Past loan performance (on-time, late, defaults)

  • Number of completed loans

  • Most predictive factor for future behavior

2. Social Trust Score

  • Connection strength between lender and borrower

  • Based on Adamic-Adar algorithm (see Social Trust Scoring)

  • Weights selective mutual friends higher

3. Loan Size Risk

  • Amount requested relative to borrower's history

  • Larger loans for new borrowers = higher risk

4. Account Quality

  • Farcaster account age and activity

  • Filters spam/bot accounts

The exact weighting and scoring formulas will be refined as we collect repayment data. Phase 0 focuses on gathering behavioral data to build a robust risk model informed by actual performance, not assumptions.


Why This Works

Research shows:

  • Prosper proved transparent risk grades work at scale [36]

  • Friend bids (capital contributions) reduce defaults by 14% [12]

  • Market-based filtering reduces defaults (high-risk loans don't fund)

Key principles:

  • Lenders make informed decisions

  • Market naturally filters high-risk loans

  • Borrowers self-select appropriate amounts

  • Short terms (30-90 days) create fast feedback loops


Security & Gaming Resistance

Risk grades incorporate quality filtering and Sybil resistance mechanisms. See Anti-Gaming & Sybil Resistance for how we defend against fake accounts and coordinated manipulation.


Lender Warnings

Before contributing, lenders see clear warnings based on risk level. View warning system β†’


Next: Social Trust Scoring Β· Anti-Gaming Β· Lender Warnings

Last updated