Gnanam ALM

What it is

Gnanam ALM is an open-source asset-projection and ALM engine for insurance regulatory work, sibling to Gnanam ESG. It projects asset cash flows and values — market and book — under scenarios, and computes regulatory results on top of that projection. Liabilities are inputs, not output: they arrive as fixed cashflow schedules produced by a separate liability model, and this engine never projects policyholder behaviour.

The first regulatory target is the Bermuda regime’s Scenario-Based Approach (SBA) to the best estimate liability. The first release is deliberately narrow: an SBA BEL engine for a fixed-income portfolio — treasury and corporate bonds, fixed and floating — plus external liability cashflows, shipped with a synthetic sample portfolio so that clone → run → BEL takes minutes.

The engine today

Complete through the SBA BEL floor, command-line, end to end — and every piece regression-locked:

• Contracts and market data. Typed input schemas and run-definition contracts; curves and conventions; the prescribed SBA scenario tables; a native reader for Gnanam ESG scenario cubes.
• Instruments. Fixed and floating bonds, callable and putable bonds priced on a Hull–White lattice, and external-cashflow assets — each with pricing and risk analytics.
• Projection. A time-grid kernel with per-currency cash accounts and scenario fan-out, and the reinvestment strategies that drive the asset–liability management itself.
• The Bermuda SBA layer. The nine prescribed scenarios, the asset-requirement solve, biting-scenario selection, the default/downgrade and transaction-cost drag, and asset-eligibility tiers with their limited-basis caps.

A four-verb command line — convert → run → inspect / compare — takes a book of vendor-extract CSVs to a sealed, reproducible SBA floor. Still to come: a graphical interface (the engine is command-line today), formatted workbook export, wider asset coverage, and the governance pack.

The standard of evidence

• Sealed run bundles. Every run is reproducible bit-identically from a single self-contained folder: inputs, outputs, diagnostics, methodology snapshot, hashes. Any third party can re-run to the same bits.
• Every number explains itself. Drill from the headline BEL down to the scenario, the timestep, the instrument, and the cash flow that produced it.
• Fail loudly. No silent clamps, no silent fallbacks, no partial result promoted to an answer.
• Provenance on every assumption. Regulation, published methodology, or judgement — tagged at the table-entry level.
• An independent oracle. The published benchmark portfolio is validated against an independently built spreadsheet model, not against the engine’s own arithmetic.

The full methodology, the validation evidence, and an honest limitations section publish with the model — the same standard the ESG is held to.

Limitations

• The v1 output is a floor, not the BEL. The first release produces a provable floor on the best estimate liability: every remaining simplification is chosen to bias the number down, and the methodology names each one.
• Liabilities are inputs. The engine never projects policyholder behaviour; liability cashflows arrive as fixed schedules from a separate liability model.
• A deliberately narrow asset universe. Fixed income at first release — treasury and corporate bonds, fixed and floating; wider coverage is post-v1 work.
• Command-line only, today. The graphical interface is still to come.

Licence and warranty

Apache-2.0, copyright Gnanam QuanTech Limited. Nothing is stubbed or held back: the open model is the complete model. No warranty of any kind. Nothing here is actuarial advice: the adopting actuary owns validation and sign-off in their own control environment, as they would for any model, purchased or built.

Take it

The repository — code, methodology, and the benchmark evidence — will be published openly when the first release ships.

Want it fitted?

If you want this engine fitted to your business — your portfolio, your liability cashflows, your reporting basis, your control framework — that fitting is the commercial work, delivered through Agaram Actuarial Limited. Write to bala@gnanamquantech.com.