Skip to main content

itofin

Idiomatic Python bindings over the libitofin Rust core, a ground-up port of QuantLib for pricing, risk, and calibration. The heavy numerics run in Rust; Python drives them. The pricing snippets below reproduce the cached Rust test oracles exactly; the abbreviated calibration snippets converge to the same targets within the tests' calibration tolerance (the full matrices live in tests/).

The API mirrors QuantLib's ql/ layout: types live in submodules (itofin.time, itofin.instruments, itofin.models, ...), while Settings and ItofinError stay at the top level.

Source API guides: currency fields, currency presets including TWD, and Black FX delta conventions. These helpers do not add currency rounding, conversion or a new option pricing engine.

Standalone constant-parameter GSR process bindings expose forward-measure transitions, not a fitted GSR model.

Pricing and process binding guides

Scope Guide Example
Ordinary bond coupon helpers BPS and ATM rate Example
Price-fed bond ATM Price-fed ATM Example
European asset-or-nothing FD pricing Asset digitals Example
European signed gap FD pricing Gap options Example
European percentage-strike FD pricing Percentage-strike options Example
Amortizing fixed-rate bonds Fixed amortization Example
Amortizing plain-Ibor bonds Ibor amortization Example
European analytic cash digitals Cash-or-nothing Example
Standalone Gaussian state process MF state Example
Standalone live-curve forward process Hull-White forward Example
Constant-notional Ibor cross-currency swaps Xccy basis Example

These helpers do not complete the wider BondFunctions overload set. The process facades are not fitted models; cash digitals cover European analytic Black-Scholes pricing, and Xccy spot is an immutable snapshot. Asset/gap FD pricing retains documented grid limitations. Amortizing constructors cover only positive, nonincreasing balances.

Install

pip install itofin

Python versions

itofin ships a single abi3 wheel that runs on CPython 3.10, 3.11, 3.12, and 3.13+. Python 3.13 is the primary, tested target; 3.10 through 3.12 are supported through the stable ABI.

Price a European option and its greeks

from itofin import Settings
from itofin.instruments import OptionType, VanillaOption
from itofin.processes import BlackScholesProcess
from itofin.time import Date, DayCounter

s = Settings()
s.set_evaluation_date(Date(15, 6, 2026))
dc = DayCounter.actual360()

process = BlackScholesProcess(60.0, 0.08, 0.0, 0.30, Date(15, 6, 2026), dc)
option = VanillaOption(OptionType.Call, 65.0, Date(15, 6, 2026) + 90, s)
option.set_engine(process)

print(f"NPV   {option.npv():.10f}")
print(f"delta {option.delta():.10f}")
print(f"vega  {option.vega():.10f}")
print(f"rho   {option.rho():.10f}")
NPV   2.1333684449
delta 0.3724827980
vega  11.3515440535
rho   5.0538998582

Price under Heston and calibrate to a flat-vol surface

Semi-analytic Heston pricing:

from itofin import Settings
from itofin.instruments import OptionType, VanillaOption
from itofin.models import HestonModel
from itofin.processes import HestonProcess
from itofin.time import Date, DayCounter

s = Settings()
s.set_evaluation_date(Date(27, 12, 2004))
dc = DayCounter.actual_actual_isda()

# HestonProcess(risk_free, dividend, spot, v0, kappa, theta, sigma, rho, reference_date, day_counter)
process = HestonProcess(
    0.0225, 0.02, 1.0, 0.1, 3.16, 0.09, 0.4, -0.2, Date(27, 12, 2004), dc
)
model = HestonModel(process)
option = VanillaOption(OptionType.Call, 1.05, Date(28, 3, 2005), s)
option.set_heston_engine(model, 64)

print(f"Heston NPV {option.npv():.10f}")
Heston NPV 0.0404774515

Calibrating the model to a flat 10% vol surface drives the vol-of-vol sigma toward zero and both v0 and theta toward the flat variance (0.10 squared = 0.01). The snippet below fits a 3-maturity by 3-strike grid; the full 21-helper matrix and its tighter tolerances live in tests/test_heston_calibration.py.

import math

from itofin import Settings
from itofin.models import CalibrationErrorType, HestonModel, HestonModelHelper
from itofin.optimization import EndCriteria, LevenbergMarquardt
from itofin.processes import HestonProcess
from itofin.termstructures import FlatForward
from itofin.time import Calendar, Date, DayCounter, Period

s = Settings()
s.set_evaluation_date(Date(15, 1, 2026))
dc = DayCounter.actual360()
ref = Date(15, 1, 2026)
risk_free = FlatForward(ref, 0.04, dc)
dividend = FlatForward(ref, 0.50, dc)
VOL = 0.10

helpers = []
for n in (1, 2, 3):
    tau = float(n)
    forward = dividend.discount(tau) / risk_free.discount(tau)  # spot = 1.0
    for moneyness in (-1.0, 0.0, 1.0):
        strike = forward * math.exp(-moneyness * VOL * math.sqrt(tau))
        helpers.append(HestonModelHelper(
            Period(n, "Years"),
            Calendar.null_calendar(),
            1.0,        # spot
            strike,
            VOL,        # 10% flat vol
            0.04,       # risk-free
            0.50,       # dividend yield
            CalibrationErrorType.RelativePriceError,
            ref,
            dc,
            s,
        ))

process = HestonProcess(0.04, 0.50, 1.0, 0.01, 0.2, 0.02, 0.5, -0.75, ref, dc)
model = HestonModel(process)
method = LevenbergMarquardt(1e-8, 1e-8, 1e-8, False)
end_criteria = EndCriteria(400, 40, 1e-8, 1e-8, 1e-8)
model.calibrate(helpers, method, end_criteria, 96)

print(f"sigma {model.sigma():.6f}")
print(f"v0    {model.v0():.6f}")
print(f"theta {model.theta():.6f}")
sigma 0.000216
v0    0.010000
theta 0.010000

Calibrate Hull-White to a swaption matrix

from itofin import Settings
from itofin.indexes import Euribor
from itofin.models import CalibrationErrorType, HullWhite, SwaptionHelper
from itofin.optimization import EndCriteria, LevenbergMarquardt
from itofin.termstructures import FlatForward
from itofin.time import Date, DayCounter, Period

s = Settings()
s.set_evaluation_date(Date(15, 2, 2002))
curve = FlatForward(Date(19, 2, 2002), 0.04875825, DayCounter.actual365_fixed())

model = HullWhite(curve, 0.1, 0.01)
index = Euribor.six_months(curve, s)

# (option tenor, swap tenor, Black vol); the full co-terminal matrix is in the tests.
swaptions = [(1, 5, 0.1148), (2, 4, 0.1108), (3, 3, 0.1070), (4, 2, 0.1021), (5, 1, 0.1000)]
helpers = [
    SwaptionHelper(
        Period(maturity, "Years"),
        Period(length, "Years"),
        vol,
        index,
        Period(1, "Years"),
        DayCounter.thirty360_bond_basis(),
        DayCounter.actual360(),
        curve,
        CalibrationErrorType.RelativePriceError,
        1.0,
    )
    for maturity, length, vol in swaptions
]

method = LevenbergMarquardt(1e-8, 1e-8, 1e-8, False)
end_criteria = EndCriteria(10000, 100, 1e-6, 1e-8, 1e-8)
model.calibrate(helpers, method, end_criteria, False)  # fix_reversion=False, fit a and sigma

print(f"a     {model.a():.7f}")
print(f"sigma {model.sigma():.7f}")
a     0.0464113
sigma 0.0057992

Price a European swaption with the Jamshidian engine

from itofin import Settings
from itofin.indexes import Euribor
from itofin.instruments import (
    EuropeanExercise,
    SettlementMethod,
    SettlementType,
    Swaption,
    SwapType,
    VanillaSwap,
)
from itofin.models import HullWhite
from itofin.termstructures import FlatForward
from itofin.time import BusinessDayConvention, Calendar, Date, DayCounter, Frequency, Schedule

s = Settings()
s.set_evaluation_date(Date(15, 1, 2026))
curve = FlatForward(Date(15, 1, 2026), 0.03, DayCounter.actual365_fixed())

fixed = Schedule(
    Date(15, 1, 2028), Date(15, 1, 2033),
    Frequency.Annual, Calendar.target(),
    BusinessDayConvention.Unadjusted,
)
floating = Schedule(
    Date(15, 1, 2028), Date(15, 1, 2033),
    Frequency.Semiannual, Calendar.target(),
    BusinessDayConvention.Unadjusted,
)
index = Euribor.six_months(curve, s)

swap = VanillaSwap(
    SwapType.Payer, 100.0,
    fixed, 0.03, DayCounter.thirty360_bond_basis(),
    floating, index, 0.0, DayCounter.actual360(), s,
)
swaption = Swaption(
    swap, EuropeanExercise(Date(15, 1, 2027)),
    SettlementType.Physical, SettlementMethod.PhysicalOTC, s,
)
swaption.set_jamshidian_engine(HullWhite(curve, 0.05, 0.01))

print(f"payer swaption NPV {swaption.npv():.10f}")
payer swaption NPV 1.5666103956

Draw random numbers as NumPy arrays

The generators under itofin.randomnumbers carry QuantLib's Python names. Sequence generators return NumPy arrays, and next_sequences(count) draws a whole (count, dimension) matrix in one call, so a path loop never crosses the binding once per path.

from itofin.randomnumbers import (
    GaussianRandomSequenceGenerator,
    UniformRandomGenerator,
    UniformRandomSequenceGenerator,
)

rng = UniformRandomGenerator(42)                       # MT19937, seeded
usg = UniformRandomSequenceGenerator(3, rng)           # 3 uniforms per sequence
gsg = GaussianRandomSequenceGenerator(usg)             # inverse-cumulative normals
print(usg.next_sequence())                             # ndarray, shape (3,)
print(gsg.next_sequences(1000).shape)                  # (1000, 3) standard normals

SobolRsg, HaltonRsg and GaussianLowDiscrepancySequenceGenerator provide the low-discrepancy counterparts with the same next_sequence / next_sequences calls.

Global bootstrap callbacks

PiecewiseYieldCurve(..., bootstrap="global") accepts keyword-only additional_penalties(times, data), additional_dates(), and additional_variables=SimpleQuoteVariables(quotes, guesses, lower_bounds). Penalties return finite residuals with a constant length during each calculation; times and data are copied lists including the reference node. Extra helpers can be repriced through helper.quote_error() inside a penalty. Additional dates and variables add unknowns and need corresponding residuals. Iterative bootstrap rejects these options.

For a futures convexity quote being fitted, pass register_conv_adj=False to its FuturesRateHelper constructor. This prevents optimizer updates from recursively invalidating the curve. The default remains observable for ordinary market quotes. See the joint futures-convexity example, checked against an independently compiled QuantLib oracle.

Callbacks run synchronously on the owning thread. Python exceptions become ItofinError from fallible pricing/bootstrap queries, including the original exception type and message. Failed queries can be retried after fixing the callback. max_date() retains its existing fallback on bootstrap failure; it neither proves a successful solve nor hides the error from a subsequent pricing query. Do not mutate quotes inside callbacks; such mutations are rejected. Callback state changes alone do not invalidate a successfully cached curve; observed market-input changes do.

Native consumers retain callbacks after the Python curve wrapper is dropped. Avoid callbacks that strongly capture the curve or an engine/model/index that retains it. Those mixed native/Python cycles need the user to break the capture; use weakref.ref(curve) for trial-curve queries. Capturing an additional helper is supported. Full mixed-graph garbage collection is tracked in #1028.

Typing checks

See typing acceptance for the pinned, bidirectional Pyright gate.

License

BSD-3-Clause, matching the libitofin core. See the repository for the layer status table, the divergences-from-QuantLib catalogue, and the Rust test oracles behind the numbers above.

Standalone plain-vanilla BlackCalculator

Immutable Black 1976 value and scalar Greeks are available through BlackCalculator. See the bounded scope and edge-case contract.

Par asset swap bindings

Plain fixed/zero bullet-bond AssetSwap supports live discounting, signed leg values, floating BPS and fair spread/clean price. Market, amortizing, callable and overnight variants are excluded. See the bounded scope and lifecycle contract.

Bounded CPI pricing

These are additive bounded routes, not CPI optionlet volatility/calibration parity.

Metadata

Release files for itofin 0.51.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for itofin 0.51.0
File Size Uploaded Provenance Actions
itofin-0.51.0.tar.gz 4.9 MB With provenance Details for itofin-0.51.0.tar.gz

Built distributions (wheels)

Table of built distributions (wheels) for itofin 0.51.0
File Actions
itofin-0.51.0-cp310-abi3-manylinux2014_x86_64.manylinux_2_17_x86_64.whl CPython 3.10 abi3 Linux glibc 2.17+ x86-64 Details for itofin-0.51.0-cp310-abi3-manylinux2014_x86_64.manylinux_2_17_x86_64.whl
itofin-0.51.0-cp310-abi3-manylinux2014_aarch64.manylinux_2_17_aarch64.whl CPython 3.10 abi3 Linux glibc 2.17+ ARM64 Details for itofin-0.51.0-cp310-abi3-manylinux2014_aarch64.manylinux_2_17_aarch64.whl
itofin-0.51.0-cp310-abi3-macosx_11_0_arm64.whl CPython 3.10 abi3 macOS 11.0+ ARM64 Details for itofin-0.51.0-cp310-abi3-macosx_11_0_arm64.whl
itofin-0.51.0-cp310-abi3-macosx_10_12_x86_64.whl CPython 3.10 abi3 macOS 10.12+ x86-64 Details for itofin-0.51.0-cp310-abi3-macosx_10_12_x86_64.whl

Total release size: 27.2 MB

Release files / itofin-0.51.0.tar.gz

Download URL itofin-0.51.0.tar.gz
Size 4.9 MB
Tags Source
SHA-256 checksum
How to use checksums
d159579630e40f54d5d0aab696eb18170086dcf7e6ad709248518cfed614569a
BLAKE2b-256 checksum
How to use checksums
a542d7fc1de13299e48f0f9410d8660d0b1e903abedaac48888997f7e307b801
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.16

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Nov 10, 2026.

Transparency log

Release files / itofin-0.51.0-cp310-abi3-manylinux2014_x86_64.manylinux_2_17_x86_64.whl

Download URL itofin-0.51.0-cp310-abi3-manylinux2014_x86_64.manylinux_2_17_x86_64.whl
Size 5.7 MB
Tags CPython 3.10 Linux glibc 2.17+ x86-64 abi3
SHA-256 checksum
How to use checksums
651a1a77618ecc2d71a26cc1716541d34946c52c3c6562f74ddd5f2035632160
BLAKE2b-256 checksum
How to use checksums
3d63915b58e774434bb2633f26c4978f7c9529fd680f91ab63bc2312ca3942e0
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.16

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Nov 10, 2026.

Transparency log

Release files / itofin-0.51.0-cp310-abi3-manylinux2014_aarch64.manylinux_2_17_aarch64.whl

Download URL itofin-0.51.0-cp310-abi3-manylinux2014_aarch64.manylinux_2_17_aarch64.whl
Size 5.7 MB
Tags CPython 3.10 Linux glibc 2.17+ ARM64 abi3
SHA-256 checksum
How to use checksums
7de6a3c7b925d9279e1f8750eee241a8160646898aeb29ef39043e5dcf46126e
BLAKE2b-256 checksum
How to use checksums
083d8a95607cdee30a5cc4b10a9c9a8b7c6064109bf029fcb64f01bac6a7a7e7
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.16

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Nov 10, 2026.

Transparency log

Release files / itofin-0.51.0-cp310-abi3-macosx_11_0_arm64.whl

Download URL itofin-0.51.0-cp310-abi3-macosx_11_0_arm64.whl
Size 5.4 MB
Tags CPython 3.10 abi3 macOS 11.0+ ARM64
SHA-256 checksum
How to use checksums
7994835aa3eeb2cafcac6a600ccc6c59d7a375ee3b1c1cd274a485b2ee2db88a
BLAKE2b-256 checksum
How to use checksums
a8921d5b0ca5744af9e56fe4d19890a3e4f0e82ac7bcde32058fc80feeb60f55
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.16

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Nov 10, 2026.

Transparency log

Release files / itofin-0.51.0-cp310-abi3-macosx_10_12_x86_64.whl

Download URL itofin-0.51.0-cp310-abi3-macosx_10_12_x86_64.whl
Size 5.4 MB
Tags CPython 3.10 abi3 macOS 10.12+ x86-64
SHA-256 checksum
How to use checksums
b89e2c86d583223a975c10735362ac7a005739f19ee7c42de9afc2399207fe7e
BLAKE2b-256 checksum
How to use checksums
ae4626d60d01947f2339195b01329e148710dd3dfa7077953a0263c746a751e0
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via twine/7.0.0 CPython/3.13.16

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Nov 10, 2026.

Transparency log

Security for itofin 0.51.0

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

Full provenance

All 5 files in this release have been uploaded with provenance metadata.

Select from 5 files

Attestation for itofin-0.51.0.tar.gz

PyPI Publish Attestation

PyPI verified that this artifact, at this checksum, originated from the publisher listed below.

Signed by GitHub Actions, verified by PyPI on Nov 10, 2026.

Transparency log

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Release history Release notifications | RSS feed

This release

0.51.0 This release

5 release files Full provenance

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page