financial-instruments-accounting

financial-instruments-accounting is an auditable, offline-capable Python library for granular financial instrument accounting calculations and reference workflows. It provides typed building blocks for classification, measurement, impairment, hedge accounting, journal preparation, disclosures and model monitoring in one reproducible package.

The methodology is designed around financial instrument accounting concepts in IFRS 9 and related presentation and disclosure interfaces. The project name is independent of the standard. The library is not affiliated with, endorsed by or certified by the IFRS Foundation or any standard setter.

Capabilities

The package exposes 80 documented public functions across five stable modules: formulas, domains, data, workflow and models.

• Scope, recognition, derecognition and continuing involvement
• Business-model and contractual-cash-flow characteristic assessments
• Classification and reclassification reference decisions
• Initial measurement, effective interest rates, amortised cost and interest revenue, including purchased or originated credit-impaired instruments
• Multi-scenario expected credit losses using PD/LGD/EAD components, discounted cash shortfalls or provision matrices
• Staging, significant-increase-in-credit-risk assessment, overlays, collateral allocation, modifications, write-offs and recoveries
• Fair-value presentation, own-credit separation and OCI rollforwards
• Hedge eligibility, effectiveness, lower-of measurement, reserve movements, recycling and basis adjustments
• Journal preparation, allowance and gross-carrying-amount rollforwards, disclosure facts and regulatory bridges
• PD, LGD, EAD and staging backtesting, challenger comparison, scenario sensitivity and audit controls
• Bitemporal input selection, strict validation, resource hashes and result lineage

Installation and first run

python -m pip install financial-instruments-accounting

The distribution name uses hyphens; the Python import namespace is financial_accounting_engine.

import financial_accounting_engine as fae

eir = fae.effective_interest_rate(1000, [(1, 1100)])
assert abs(eir - 0.1) < 1e-9

dataset = fae.load_reference_data("SMALL_SA_RETAIL_FIA")
result = fae.run_dataset(dataset)
assert result.status == "APPROVED_REFERENCE"
print(fae.get_metrics(result))

The command-line interface uses the same public workflow:

financial-instruments-accounting doctor
financial-instruments-accounting profiles
financial-instruments-accounting reference SMALL_SA_RETAIL_FIA --output ./results

Reference data and reproducibility

Two complete synthetic profiles are bundled for offline examples, integration tests and cross-implementation comparison. Each profile contains 19 sealed XLSX input workbooks plus canonical JSON data. No client, bank or personal data are included.

Reference calculations are checked against independently specified formula cases and frozen full-profile results. The repository also contains API-contract tests, property tests, correction evidence, resource manifests and installed wheel/sdist smoke tests. GitHub Actions verifies Python 3.10 through 3.13.

Scope and legal notice

APPROVED_REFERENCE means that the package's automated reference controls have passed. It is not an institutional approval, an audit opinion or an accounting conclusion. Users remain responsible for contract interpretation, accounting policy choices, approved models and parameters, market data, materiality, governance, ledger integration and financial-statement review.

The software does not implement every requirement, exception or disclosure in IFRS Accounting Standards and is not a substitute for the standards or for professional accounting, audit, legal or risk advice. Official source metadata and links are provided for traceability; standard texts and other third-party materials are not redistributed. Obtain authoritative, current materials from their publishers and observe their terms.

The Apache License 2.0 applies to this software and its synthetic project data. It does not grant rights in third-party standards, publications or trademarks. See DISCLAIMER.md, THIRD_PARTY_NOTICES.md and NOTICE in the source repository.

Documentation and project links

• Source repository
• API documentation and tutorials
• Issue tracker
• Security policy
• Imprint
• Privacy policy

Copyright 2026 RiskDataScience GmbH. Maintainer contact: riskdatascience@web.de.