abaquant

AbaQuant

AbaQuant is an applied actuarial and quantitative-finance library for Python. Version 1.0.0rc1 stabilizes the public namespace around derivatives, financial mathematics, market data, credit analytics, portfolio construction, risk dashboards, visualization, reports, rate curves, and provenance-aware workflows.

Installation

pip install abaquant

For a local source checkout:

python -m pip install -e .

Optional extras are grouped by workflow:

python -m pip install -e .[market]
python -m pip install -e .[visual]
python -m pip install -e .[dev]
python -m pip install -e .[docs]

AbaQuant targets Python 3.11 through 3.14; the release workflow validates each version before publication. Live market-data workflows and visualization backends are optional; the mathematical core depends only on NumPy, pandas, and SciPy.

Minimal examples

Derivatives

from abaquant.derivatives import BlackScholesMertonModel

model = BlackScholesMertonModel(
    spot_price=100.0,
    strike_price=105.0,
    maturity_years=1.0,
    risk_free_rate=0.04,
    volatility=0.20,
)

price = model.price("call")
greeks = model.greeks("call")

Market data and option-chain analytics

from abaquant.marketdata import get_ticker

ticker = get_ticker("AAPL")
spot = ticker.spot()
# Live data requires an installed provider extra and provider availability.

Portfolio construction

from abaquant.portfolio import PortfolioAllocator

allocator = PortfolioAllocator.from_returns(returns)
weights = allocator.max_sharpe()
report = allocator.report()

Rates

from abaquant.rates import ManualRateProvider, get_rate_curve

curve = get_rate_curve(
    provider=ManualRateProvider({1.0: 0.045, 5.0: 0.047, 10.0: 0.049})
)
discount = curve.discount_factor(5.0)

Provenance

from abaquant.core import DataProvenance

provenance = DataProvenance(
    provider="manual",
    dataset="example",
    request={"symbol": "AAPL"},
)
metadata = provenance.as_dict()

Stable v1 namespace map

The root abaquant namespace re-exports the documented public facades and defines abaquant.__version__.

Live-data warnings

Market-data providers are optional and lazy. Object construction should not make network requests unless a retrieval method explicitly asks for provider data. Provider data can be unavailable, incomplete, stale, adjusted, restated, or subject to provider-specific terms and rate limits.

For SEC EDGAR requests, set a project-specific contact user agent when using live data:

export ABAQUANT_SEC_USER_AGENT="your-app/1.0 your.email@example.com"

For FRED requests, provide a FRED API key through the provider constructor or environment where supported.

Model assumptions

AbaQuant implements educational and research-grade quantitative routines. Major assumptions include:

• Black-Scholes-Merton assumes lognormal dynamics, constant volatility, frictionless markets, and continuous trading.
• Black-76 uses forward pricing conventions and constant lognormal forward volatility.
• Bachelier uses normal price dynamics and can accommodate negative underlyings or rates.
• Heston, SABR, Merton jump diffusion, NIG, and Variance-Gamma models rely on numerical approximations and calibration quality.
• Portfolio optimizers are sensitive to estimated returns, covariance matrices, constraints, sampling frequency, and transaction-cost assumptions.
• Credit proxy metrics are accounting-derived heuristics, not agency ratings.
• Gaussian copula credit simulation depends materially on default probabilities, recoveries, and correlation assumptions.
• Backtests are historical simulations; they are not forecasts.

Not investment advice

AbaQuant is not an investment adviser, broker, credit-rating agency, or risk-management system. Outputs are model-derived estimates for research and education, not trading, lending, investment, tax, accounting, or legal advice.

Development checks

python -m pip install -e ".[dev,docs,market,visual]"
python -m ruff format --check .
python -m ruff check .
python -m pytest
python scripts/check_documentation.py
python -m sphinx -W --keep-going -b html docs docs/_build/html
python -m build
python -m twine check dist/*

See CONTRIBUTING.md for the complete contributor workflow and the hosted documentation for API contracts, model assumptions, and examples.

Documentation source layout

The Sphinx documentation source is written in native reStructuredText under docs/ and organized into topic subfolders: getting-started/, reference/, domains/, operations/, development/, and api/. The api/ tree is generated from the source package by python scripts/generate_api_docs.py; it mirrors all 109 Python modules/packages and renders detailed autodoc entries for the public callable surface.