Skip to content

Releases: Aniket2002/rates-engine-cpp

v2.3.0 - SOFR hedging, durable lifecycle and portable Windows binaries

Choose a tag to compare

@Aniket2002 Aniket2002 released this 27 Oct 16:57

v2.3.0: SOFR hedging, durable lifecycle and portable Windows binaries

This release connects computed sensitivities to a constrained hedge decision,
adds complete contractual payment checks and durable lifecycle journals, and
removes compiler-DLL PATH requirements from default Windows executables.

  • A 5Y SOFR hedge study sizes 2Y/4Y/6Y positions under USD 1.5 million per-instrument
    limits and USD 1,000 lot rounding. Quote-DV01 norm falls 51.68%; assumed entry
    costs are USD 379.73. Ten full-repricing shocks show residual and nonlinear risk,
    including cases that worsen. QuantLib independently checks 44 instrument PVs
    and 28 quote sensitivities. JSON, CSV and PNG/SVG results are committed.
  • C++ exposes the full lag-adjusted SOFR payment schedule. The basis workflow
    derives final obligations from realized coupons and reconciles the complete
    inventory and amounts before normal-maturity closure.
  • An optional Python/SQLite journal wrapper serializes concurrent writers,
    atomically validates/appends batches across streams, rejects conflicting
    retries and replays persisted state. Strict contract inventories prevent
    missing-payment closure. Tests include four processes, 80 concurrent events
    and process death before commit.
  • Firm OIS import policy adds execution-size, expiry and source/usage-reference
    checks. Actual executable quotes still require authorized external data;
    synthetic examples remain clearly labeled.
  • Windows builds embed compiler runtimes by default. A PowerShell launcher
    builds/tests/runs the example, and CPack produces a CLI/library ZIP with example
    inputs, build information and runtime license notices.
  • README encoding is repaired, committed terminal auto-approval is removed,
    and implementation walkthroughs explain calibration, accrual and collateral.

Run

From a configured source checkout:

.\run.ps1 -WithPython -Test

Or download and extract rates-engine-2.3.0-Windows-AMD64.zip, open PowerShell in
its extracted root and run:

.\bin\rates_cli.exe --quotes examples/synthetic_quotes.csv --swap examples/sample_swap.csv --out results/demo

The ZIP contains the native CLI tools and static library. Python workflows are
built from source for the chosen interpreter. The Windows ZIP targets x64 with
the Universal CRT; its C++ library uses the LLVM-MinGW ABI.

Verification

Local Windows x64, LLVM-MinGW Clang 23.1.3, Python 3.14.0, QuantLib 1.44:

Check Result
Release, bindings enabled 100/100 CTest tests passed
Debug, bindings enabled 100/100 CTest tests passed
Full Release suite with system-only PATH 100/100 passed
Independent hedge prices and quote risks 72 comparisons passed
Durable lifecycle and contract checks 20 passed
Hedge solver fixtures / market-data unit tests 4 / 7 passed
Installed C++ consumer with system-only PATH Passed
Extracted ZIP, relocated CLI tools and actual pricing Passed
C++ formatting and clang-tidy build Passed

See runtime evidence,
hedge results, and
durable lifecycle results.
The cross-platform workflow runs the new tests and requires independent
QuantLib hedge validation. The release commit's
22 October CI run
passed all seven jobs: Linux GCC Release, Linux Clang Debug, Windows MSVC Release,
macOS Release, sanitizers, static analysis and LLVM coverage.

This remains a deterministic swaps project. Options, stochastic models, XVA,
licensed executable feeds and a production clearing service are not claimed.

v2.2.0 - SOFR lifecycle pricing and auditable market data

Choose a tag to compare

@Aniket2002 Aniket2002 released this 21 Oct 18:29

v2.2.0 — SOFR lifecycle pricing and auditable market data

SOFR swaps can now mix realized and forecast overnight observations, price
seasoned and matured/unpaid coupons, and exclude settled cash flows. New APIs
support lookback, observation shift, lockout, spot settlement and payment lags,
accrued/clean PV, lag-aware OIS calibration and separate discount/projection risk.

The New York Fed adapter captures actual public SOFR fixings with immutable
raw-response archives, explicit local availability and revision history. The
OIS quote importer checks source, timestamps, units, calendar/conventions,
required tenors, bid/ask bounds, freshness and configurable rate/spread limits.
An offline example combines public fixings with clearly marked synthetic quotes.

Validation: 86 Release tests, 82 Debug tests, 6,892 independent lifecycle and
calibration comparisons, and 28 captured-fixing cash-flow/risk comparisons pass.
Formatting, clang-tidy, coverage and the installed consumer were checked.
An observation-period normalization defect was corrected during external validation.

The changes are split into focused commits. Existing v2.1 APIs keep their scope.
Public fixing capture does not provide executable OIS curves: those still require
an external quote provider. Contract amendments, novations, clearing, collateral
and production monitoring remain outside the implementation. See the
lifecycle guide
and verification record.

v2.1.0 - OIS, multi-curve valuation and validated risk reporting

Choose a tag to compare

@Aniket2002 Aniket2002 released this 21 Oct 17:51

v2.1.0 — OIS, multi-curve valuation and validated risk reporting

Continuous-zero quote risk now reports the exact calibrated maturity, including
weekends and holidays. Calibration and reporting share one pillar-date rule;
regressions check direct/shared risk and CLI/batch exports.

The library and Python bindings add ACT/360 SOFR compounding, annual OIS
calibration/pricing, separate discount and forwarding curve calibration, and
quote risk that recalibrates both curves after discount shocks. Existing
single-curve calls retain their behavior. Independent QuantLib OIS/swap helpers
validate 230 new comparisons across three synthetic cases.

The README follows one worked swap from quotes to PV and risk, with three
reproducible PNG/SVG charts. A model-validation case study explains seven PV
shock scenarios and their nonlinear residuals.

Verification: 73 Release tests, 71 Debug tests, 1,285 original numerical
comparisons, 2,193 convention comparisons and all new multi-curve comparisons
pass locally. The full Fed replay completes 24,900 valuations without failed
trades. Library line coverage is 98.95%; formatting, static analysis and the
installed consumer pass. See the verification record.

OIS scope: fully future annual ACT/360 coupons, explicit calendars and zero
payment/settlement lag. No observation shifts, lookbacks, lockouts or mixed
realized/forecast coupons. Existing CSV tools remain single-curve.
See the extension guide for assumptions and reproduction.

All seven cross-platform CI jobs passed: verification run.

The evidence archive contains the three charts in PNG/SVG, numerical comparisons, replay results, source hashes and CI status. Verify it against SHA256SUMS.txt.