ModelStack finance models, rebuilt with institutional bones

Twenty-two finance products, rebuilt in one pass

Today we shipped a full rebuild of every finance product in the ModelStack catalog. Twenty-two items, one deploy. Twelve spreadsheets (LBO, merger, M&A valuation, three flavors of three-statement, cap table, VC fund returns, freelancer P&L, ecommerce unit economics, startup valuation calculator). Six documents (pitch deck brief, sell-side data room checklist, buy-side due-diligence checklist, IC memo, PPM, monthly close SOP). The bundles got the same treatment.

The driver was installing the claude-for-financial-services plugin family on the build side. Nineteen plugins, each carrying spec conventions for one corner of the finance stack: LBO mechanics, merger accounting, three-statement linkage, waterfall structures, due-diligence severity tags. We pointed the regeneration pipeline at those conventions and let it rewrite every file from the ground up. Old versions are gone from the store. New ones ship today.

The reason to read this post even if you're not buying anything: the work surfaced a clean list of the gaps that separate institutional-grade models from the lookalike templates that flood the internet. The rest of this is that list.

Most finance templates are visual mockups

Open any of the top-ranked Excel templates on Etsy, Gumroad, or a search-result PDF. The cover sheet is beautiful. The fonts match. There's a chart. Then you change one input and three things break.

This happens because the seller built the template to photograph well, not to hold up. The formulas reference the wrong sheets. The interest expense doesn't iterate. The balance sheet plugs to zero with a hardcoded number. The exit IRR uses an MOIC approximation. None of this matters until you actually use the file, and by then you've already paid.

The rebuild was about closing that gap. Three things changed across every model.

One: a Checks tab on every spreadsheet

Every XLSX in the catalog now has a tab named Checks. It runs a series of TRUE/FALSE integrity tests. Conditional formatting paints green for pass, red for fail. You can tell from glancing at one cell whether the model holds together.

What the LBO Checks tab asks: does the sources line equal the uses line? Does total debt at year five match the sum of revolver, senior, and sub balances? Does ending cash on the balance sheet match ending cash on the cash flow statement? Does the IRR computed from cash flows match the IRR computed from equity proceeds? Nine of these on the LBO alone.

The three-statement models (SaaS, restaurant, startup pitch, SaaS fundraising) get a per-year balance check. Assets minus liabilities minus equity should equal zero in every column. If it doesn't, you see red. If it does, you can present the model to a committee without a private rehearsal first.

None of this is technically hard. It's just unusual to include. The reason most templates don't have it is that the seller doesn't actually run the checks while building, so they wouldn't pass them if forced. Adding the tab forced us to make every model pass.

Two: the small details that get fixed in week two of an associate program

There's a list of details every banking analyst learns in their first two months that almost no for-sale template includes. The rebuild forced us to handle them.

The LBO uses begin-balance interest. Interest expense in year N is calculated off the year N minus 1 debt balance, not the average or the ending balance. This avoids the canonical iterative-calculation circular reference that breaks the model when you change scenarios. If you've ever opened a template and seen "Enable iterative calculations? [Y/N]" pop up, you've met a model that didn't do this.

The merger model carries full intangible amortization. When the deal premium creates goodwill, the rebuild splits it correctly between goodwill (not amortized) and identifiable intangibles like customer relationships and developed technology (amortized over a defined useful life). The amortization flows through the income statement, hits deferred tax, and reverses on the balance sheet. Most templates skip this and shove the entire premium into goodwill, which inflates pro-forma EPS by a wide margin.

The three-statement variants tie. The income statement flows to retained earnings. The cash flow statement reconciles net income to ending cash. Working capital changes on the cash flow match the year-over-year change in balance sheet accounts. The plug is zero in every column without forcing it.

The VC fund model uses a proper European-style waterfall. Limited partners get cap return first. Then an 8% hurdle. Then a general partner catch-up tier. Then 80/20 above the catch-up. Twelve portfolio companies, each with its own exit timing and multiple. The carry math works on actual cash flows, not a single aggregate multiple at the fund level.

The cap table walks from incorporation through Series D with option-pool refreshes at each round and a full exit waterfall at the end. Liquidation preferences, anti-dilution, the works.

Three: formula colors, so models are auditable at a glance

Every cell in every spreadsheet now follows the same color convention. Blue is a hardcoded input. Black is a formula on the current tab. Purple is a same-tab link. Green is a cross-tab link.

This is a banking analyst standard, and it changes how fast you can audit a model. You inherit a file. You want to know what assumptions drive a number. You click the cell and read the color. Blue means change me. Green means trace me back to another tab. Black means the math happens here.

Without color conventions, every audit starts with "open the formula bar and squint." With them, the model tells you where to look.

What changed on the document side

The DOCX files got the same treatment from a different angle. The pitch deck brief now has thirteen slides, each with one question, what to include, and common mistakes. The data-room checklist has eight folders and over seventy line items, each tagged CRITICAL, NICE, or NEVER. The buy-side due-diligence checklist has eight workstreams with severity tags. The IC memo template has twelve sections in committee-ready order. The PPM template has fourteen. The monthly close SOP runs T+10 day-by-day with an owner matrix attached.

The thing the documents share with the spreadsheets: they encode opinions, not blank fields. "Put your numbers here" is not a template. "Here are the seven things a buyer's diligence team will ask, ranked by severity, with examples of what bad answers look like" is a template.

The takeaway for anyone shopping for finance models

If you're evaluating any spreadsheet template (ours or anyone else's) before you buy, the test is simple. Ask for the Checks tab. Ask whether the income statement, balance sheet, and cash flow tie in every column. Ask how interest is computed in the LBO. Ask whether the merger model amortizes identifiable intangibles separately from goodwill. Ask what color the inputs are.

If the seller can't answer those questions, the model is a mockup. You can build a mockup yourself in an afternoon. You're paying for the parts that take a week to get right, not the parts that take an hour to format.

The rebuilt ModelStack finance catalog is live now. The XLSX files are at the same URLs they were yesterday with a new version inside. If you bought one of these in the last twelve months, email hello@modelstack.digital and we'll send you the rebuilt version free.