Build auditable models and schedules

Build the adjustment schedules twice: a formatted workbook for cell-by-cell review and a dependency-free Python script for readable recomputation. Both consume the same labeled inputs and produce the same control totals.

The workbook separates instructions, inputs, calculations, outputs, and checks. Input cells have units, dates, source IDs, owners, and assumption states. Formulas use stable references and visible subtotals. Error checks are prominent and print cleanly. Do not hide logic in merged cells, named ranges nobody can inspect, or a single formula spanning several accounting decisions.

The Python version should read like a careful notebook without external dependencies. Use named variables, small functions, explicit sign conventions, and formatted outputs with commas, parentheses, currency symbols, and percentages. Print each major bridge and check. A learner should be able to change one assumption near the top and understand the result.

Reperform the base case, then change one input deliberately and explain every downstream effect. If the two formats disagree, stop. The model is not a voting system; locate the difference and preserve the correction in the revision trail.

The model proves arithmetic under supplied inputs. It does not supply the recognition conclusion, estimate support, or authorization that owns those inputs.