Introducing Credkit

I got tired of fighting with QuantLib to model simple mortgages. So I built something easier.

The Problem

QuantLib is great for complex derivatives. But for simple amortizing consumer loans it’s complete overkill. You spend more time wrestling with schedule builders and leg constructors than actually modeling the loan.

These are relatively simple instruments, so they should be similarly simple to model

The Solution

from credkit import Loan, Money, InterestRate

loan = Loan.mortgage(
    principal=Money.from_float(300000.0),
    annual_rate=InterestRate.from_percent(6.5),
    term_years=30,
)

payment = loan.calculate_payment()  # ~$1,896.20/month
schedule = loan.generate_schedule()  # Done.

That’s it. No ceremony, no fighting with abstractions designed for exotic swaps.

What You Get

Domain-specific primitives:

• Money, InterestRate, Spread types that actually make sense
• Day count conventions, payment frequencies, business day calendars
• Full amortization support (level payment, level principal, interest-only, bullet)

Type-safe and immutable:

• Frozen dataclasses with full type hints
• Your IDE will actually help you
• No accidental mutations

Cash flow modeling:

• Generate schedules with principal/interest breakdown
• Discount curves for NPV calculations
• Filter and aggregate however you need

Status (v0.2.0)

148 passing tests. Works end-to-end: create a loan → generate schedule → calculate NPV.

Currently USD-only, focused on US consumer lending (mortgages, auto, personal loans).

What’s Next

• Credit risk metrics (PD, LGD, ratings)
• pandas integration
• Prepayment modeling

Try It

pip install credkit

GitHub: github.com/jt-hill/credkit
PyPI: pypi.org/project/credkit License: AGPL-3.0 (other commercial options available)

Feedback?

This is early. If you work in or near credit and have ideas or pain points, let me know. What would make your life easier?