Skill: Calculate Loan Default Risk

Domain

financial_services

Description

Calculates probability of default (PD) for loan applications using proprietary credit scoring models that incorporate traditional and alternative data sources.

Tags

banking, credit-risk, loan-underwriting, default-prediction, financial-services

Use Cases

• •Loan application decisioning
• •Portfolio risk assessment
• •Pricing risk adjustment
• •Regulatory capital calculation

Proprietary Business Rules

Rule 1: Scorecard Weighting

Proprietary weights for credit factors based on historical default correlation analysis.

Rule 2: Segment-Specific Models

Different scoring models for prime, near-prime, and subprime segments.

Rule 3: Behavioral Overlay

Recent payment behavior adjustments to base credit score.

Rule 4: Concentration Limits

Industry and geographic concentration risk adjustments.

Input Parameters

• •application_id (string): Loan application identifier
• •credit_score (int): FICO or equivalent score
• •debt_to_income (float): DTI ratio
• •loan_amount (float): Requested loan amount
• •loan_purpose (string): Purpose of loan
• •employment_months (int): Months at current employer
• •annual_income (float): Annual gross income
• •existing_debt (float): Total existing debt
• •payment_history (dict): Recent payment behavior

Output

• •probability_of_default (float): PD score (0-1)
• •risk_grade (string): Internal risk grade
• •decision (string): Approve, decline, or refer
• •pricing_adjustment (float): Rate adjustment bps
• •risk_factors (list): Contributing risk factors

Implementation

The risk calculation logic is implemented in risk_calculator.py and references scoring parameters from CSV files:

• •segments.csv - Reference data
• •adjustments.csv - Reference data
• •loan_purposes.csv - Reference data
• •risk_grades.csv - Reference data
• •pricing_grid.csv - Reference data
• •decision_thresholds.csv - Reference data
• •parameters.csv - Reference data.

Usage Example

from risk_calculator import calculate_default_risk

result = calculate_default_risk(
    application_id="APP-2024-12345",
    credit_score=720,
    debt_to_income=0.35,
    loan_amount=25000,
    loan_purpose="debt_consolidation",
    employment_months=36,
    annual_income=85000,
    existing_debt=15000,
    payment_history={"late_30_count": 0, "late_60_count": 0}
)

print(f"PD: {result['probability_of_default']}")
print(f"Decision: {result['decision']}")

Test Execution

from risk_calculator import calculate_default_risk

result = calculate_default_risk(
    application_id=input_data.get('application_id'),
    credit_score=input_data.get('credit_score'),
    debt_to_income=input_data.get('debt_to_income'),
    loan_amount=input_data.get('loan_amount'),
    loan_purpose=input_data.get('loan_purpose'),
    employment_months=input_data.get('employment_months', 0),
    annual_income=input_data.get('annual_income'),
    existing_debt=input_data.get('existing_debt', 0),
    payment_history=input_data.get('payment_history', {})
)