open-riskpy 0.2.6

C++ powered actuarial engine: FactorModel, Monte Carlo, Loss Triangle, Fourier aggregate loss, and batch Excel export.

RiskPY: Enterprise Actuarial Engine 🚀

A hyper-fast, C++ powered pricing framework for actuaries (P&C, Life, Health): declarative rating, Monte Carlo, chain-ladder reserving, Fourier aggregate loss, batch Excel export, and an optional desktop GUI.

pip install open-riskpy

from riskpy import FactorModel, FourierTransform

model = FactorModel(1000.0)
model.add_multiplier("state", "FL", 3.0)
print(model.calculate({"state": "FL"}))  # 3000.0

pmf = FourierTransform.compound_poisson_pmf([0.0, 1.0], expected_frequency=2.0, grid_size=64)

v0.2.6 — FourierTransform, hot-path performance, headless-safe imports, wheels via GitHub Actions → PyPI. See RELEASE_NOTES.md.

---

Install (end users)

pip install open-riskpy
# upgrade
pip install -U open-riskpy

# Optional GUI charts (Monte Carlo plots)
pip install "open-riskpy[gui]"

Pre-built wheels are published for common platforms. From source you need a C++17 compiler and CMake (pulled automatically by the build backend when missing).

Core install has no heavy Python deps (matplotlib is optional via [gui]).
import riskpy is headless-safe (no Tkinter required). The GUI (UnderwritingApp) loads only when you use it.

---

Install (developers)

git clone https://github.com/slimboi34/RiskPY.git
cd RiskPY
make install    # pip install -e ".[dev]"
make test       # pytest
make smoke      # quick import check
# or: make check

Without Make:

pip install -e ".[dev]"
pytest tests/

---

Core features

Module	Role
FactorModel	Multiplicative rating rules (exact + numeric bands)
MonteCarloSimulator	P&C, life, health, GBM, catastrophe simulations
ActuarialMath	PV, FV, loss ratio, mortality
LossTriangle	Chain ladder, IBNR
ExperienceRating / ExposureRating	Credibility, ILFs, layers
RateAnalyzer	On-level, trend, combined ratio, indicated rate change
FourierTransform	FFT / iFFT, convolution, compound-Poisson PMF
UnderwritingApp	Tkinter GUI + CSV batch → Excel
RiskEngine / ExcelExporter	Low-level orchestration & xlsx

Docs: docs/API_REFERENCE.md · docs/TUTORIALS.md · docs/PUBLISHING.md

---

Quick examples

Factor pricing

from riskpy import FactorModel

model = FactorModel(initial_base_rate=1000.0)
model.add_multiplier("state", "FL", 3.0)
model.add_numeric_band_multiplier("age", 16, 25, 2.0)
premium = model.calculate({"state": "FL", "age": 19.0})  # 6000.0

Fourier compound Poisson (no sampling error)

from riskpy import FourierTransform

# Severity always 1 → aggregate is Poisson(λ)
pmf = FourierTransform.compound_poisson_pmf(
    severity_pmf=[0.0, 1.0],
    expected_frequency=2.0,
    grid_size=64,
)

Monte Carlo

from riskpy import MonteCarloSimulator
import numpy as np

sim = MonteCarloSimulator(trials=100_000)
losses = sim.simulate_aggregate_loss(5.0, 10.0, 1.5)
print(np.percentile(losses, 99))

Batch CSV → Excel (needs Tk/desktop for the app wrapper)

from riskpy import UnderwritingApp, FactorModel

app = UnderwritingApp(title="Auto Rater", excel_template="template.xlsx")
app.add_field("state", "State", "A", choices=["NY", "CA", "FL"])
app.add_field("age", "Age", "B")
app.set_premium_column("C", "Premium")
app.set_factor_model(FactorModel(100.0))
total, n = app.calculate_batch("book.csv", "quotes.xlsx")

---

Releasing to GitHub + PyPI (maintainers)

One-time: configure PyPI Trusted Publishing for this repo (publish.yml, environment pypi). Details in docs/PUBLISHING.md.

# After merging to main and bumping version in pyproject.toml:
make release
gh release create v0.2.6 --generate-notes
# Actions → "Publish to PyPI" builds wheels + sdist and uploads

Workflow	When	What
CI	every push/PR to main	build + pytest (Linux/macOS)
Publish	GitHub Release / manual	wheels + sdist → PyPI

---

Building from source (details)

Tool	Notes
C++17 compiler	Apple Clang, GCC ≥ 9, MSVC 2019+
CMake ≥ 3.15	Injected by the build backend if needed
Python ≥ 3.8	3.10+ recommended
Ninja / ccache	Optional; speeds CI and local rebuilds

export CMAKE_BUILD_PARALLEL_LEVEL=$(sysctl -n hw.ncpu 2>/dev/null || nproc)
pip install .
pytest tests/

---

License

MIT — free for personal and commercial use.

• GitHub: github.com/slimboi34/RiskPY
• PyPI: pypi.org/project/open-riskpy