Add Inventory Modelling: Four Methods (monthly), and move model links to the apex domain

[anthomakr]

Two changes: a new model page, and a domain sweep the repo had been missing since the apex migration.

New model: Inventory Modelling: Four Methods (monthly)

Open in Layerz · models/inventory-methods/

Four ways to forecast inventory, computed side by side on the same business over 72 months, each in its own tab. One Inventory method switch decides which one feeds working capital, cash flow and the balance sheet. Nothing downstream references a method directly.

Method	Dec 2025	Dec 2029	Implied DIO 2029
A. Percent of COGS	4,315,200	6,849,537	68
B. Inventory days	4,322,192	6,255,285	62
C. Unit build	4,330,667	5,984,138	59
D. Ageing from flows	3,509,000	6,051,723	60

A, B and C are calibrated to agree within 0.4% in 2025 and are 865,398 apart by 2029, because a fixed ratio cannot express an efficiency gain: method A's implied days never move.

Why monthly. The model was originally yearly. It could not express stock ageing at all: on flat annual flows a "last 180 days of purchases" rule is just a DIO of 180 in disguise. 2024 is a ramp-in year that gives the trailing-12-month windows a full history and absorbs the working capital build; results are read from 2025.

What method D adds. It forecasts purchases and lets the stock follow, rather than inferring the stock from COGS. Under FIFO the remaining stock is always the most recent purchases, so age is a subtraction rather than a layer simulation:

Stock older than 60 days  = MAX(0, inventory − purchases of the last 2 months)
Write-off (over 180 days) = MAX(0, stock before write-off − purchases of the last 6 months)

The result is the point of the whole thing: at normal purchasing the write-off never fires, and the stock ages anyway, from 16% to 22% over 60 days. A business turning stock in 60 days never holds anything for 180, so the accounting alarm stays silent for years while a quarter of the warehouse quietly gets old. No ratio method can produce that column.

Balance check verified at 0 on all 72 periods and all four methods, re-run with the switch on each value. Annual P&L totals and the December inventory figures reproduce the earlier yearly build to the euro.

Apex domain sweep

layerz.cc became the canonical origin on 2026-08-06 and app.layerz.cc now 308s to it. Every Open in Layerz button, OG card and workflow URL in this repo still pointed at app., so each one took a redirect hop.

More pressing: check-links.yml was about to start failing. It asserts a bare 200 and app.layerz.cc/models/<id> answers 308. Its last green run was 2026-08-03, three days before the migration, so the next scheduled run would have failed on all 12 models. Fixed here and replayed: 13 models, 26 URLs, all 200.

The workflow now also runs on pull requests, so a dead or mistyped model link is caught before merge instead of by the next Monday cron. The id-collection grep accepts both forms, so a stray app. link in a future page still gets checked rather than silently dropped.

Notes

• The branch was rebased onto origin/main first: it carried four commits that were content-identical duplicates of main's, which the rebase dropped.
• The model is promoted to a Layerz template at user scope, not system. Per CLAUDE.md this catalog and the public template catalog are reconciled deliberately, not assumed to match.
• The earlier yearly model is superseded and no longer referenced anywhere in the repo.

🤖 Generated with Claude Code