VelesQuant

VelesQuant is a high-performance financial valuation library written in C++ and exposed to Python using pybind11. It provides tools for pricing complex financial derivatives using advanced models and numerical techniques.

Quick Start

VelesQuant provides high-level Python wrappers in velesquant.models for ease of use.

import numpy as np
from velesquant.models import SabrModel, HestonModel, HullWhiteModel
from velesquant.market.curves import DiscountCurve

# --- SABR Volatility Model ---
sabr = SabrModel(maturity=1.0, forward=100.0, beta=0.5)

# Calibrate to market data
strikes = [90.0, 100.0, 110.0]
quotes = [0.25, 0.20, 0.22]  # Implied volatilities
sabr.calibrate(strikes=strikes, quotes=quotes, calibration_target="Volatility")

print(f"Calibrated SABR: alpha={sabr.alpha:.4f}, nu={sabr.nu:.4f}, rho={sabr.rho:.4f}")

# Get implied volatility for a strike
iv = sabr.implied_vol(105.0)

# --- Heston Stochastic Volatility ---
heston = HestonModel(spot=100.0, var0=0.04, kappa=2.0, theta=0.04, xi=0.3, rho=-0.7, seed=42)

# Price a vanilla option
price = heston.price_option(maturity=1.0, forward=100.0, strike=100.0, option_type="call")

# --- Hull-White Interest Rate Model ---
# 1. Define a Discount Curve
curve = DiscountCurve(
    times=[0.0, 1.0, 5.0, 10.0],
    dfs=[1.0, 0.95, 0.80, 0.65]
)

# 2. Initialize Model
hw = HullWhiteModel(kappa=0.1, sigma=0.01)

# 3. Price a Swaption
from velesquant.instruments.rates import Swaption
swaption = Swaption(expiry=1.0, tenor=5.0, strike=0.03, pay_frequency=0.5)
price = hw.price(swaption, curve)

# 4. Simulate Short Rate (Vectorized)
times = np.array([0.25, 0.5, 0.75, 1.0])
paths = hw.simulate(times, curve)

Available Models

For a comprehensive list of supported financial instruments and their pricing methods, please refer to Supported Instruments.

Performance

VelesQuant is built for extreme performance:

• C++20 Core: Leveraging modern language features for safety and speed.
• SIMD Optimized: Critical numerical paths are designed for vectorization.
• Multi-threaded: Parallel PDE solvers and Monte Carlo simulations utilizing OpenMP.
• QuantLib Integration: Built on top of the industry-standard QuantLib library.

Testing

Python Tests

The recommended way to develop is using uv for lightning-fast environment management:

uv sync --all-extras
uv run pytest

Alternatively, using standard pip:

pip install ".[dev]"
pytest

If using poetry:

poetry run pytest

C++ Tests

./build/tests_cpp/velesquant_tests

Contributing

We welcome contributions! Please follow these steps:

1. Bug Reports: Open an issue describing the bug and providing a reproducible example.
2. Feature Requests: Use the issue tracker to propose new features.
3. Pull Requests:
   • Fork the repository.
   • Create a feature branch.
   • Ensure all C++ and Python tests pass.
   • Submit a PR with a clear description of changes.

License

MIT License