Valuation Engine Architecture

From ticker to board deck in one artifact

7 companies. 10 assumptions. 5-year pro-forma model. DCF, reverse DCF, AI equity research. Export to Excel with real formulas, PPT formatted, PDF print-ready. One living document.

Valuation engine data flow

01 — The Pipeline

Five stages. One artifact.

Every number you see in the Pro-Forma Valuation Engine flows through this pipeline. Change one assumption — every view and every export updates instantly.

Input

Ticker Data

7 companies built-in (AAPL, MSFT, GOOGL, NVDA, AMZN, META, TSLA). Search to load. Revenue, COGS, OpEx, D&A, balance sheet, shares outstanding — all pre-loaded from SEC filings.

Configure

10 Assumptions

Revenue growth, gross margin, OpEx ratio, tax rate, WACC, terminal growth, buyback %, safety margin, discount rate, share dilution. Every slider recomputes the entire model in real time.

Compute

Valuation Engine

5-year pro-forma income statement, balance sheet, and cash flow statement. DCF valuation with terminal value. Reverse DCF ("What's Priced In"). All computed client-side — no server round-trip.

Analyze

4 Views

Overview (KPIs, charts, valuation rating). 3-Statement Model (full financial tables). What's Priced In (reverse DCF sensitivity). Equity Research (AI-generated analysis via Claude).

Export

3 Formats

Excel with live formulas (ExcelJS). PowerPoint fully formatted with charts (PptxGenJS). PDF print-ready (jsPDF). All generated client-side from the same model state.

02 — What's Next

Round-trip editing.
Coming soon.

Today the engine exports real files — Excel with formulas, PPT with slide masters, PDF with proper typography. Those exports are first-class citizens in their native formats.

The next step: when your CFO edits the Excel assumptions or your VP restructures the PowerPoint deck, those changes can feed back into the model to create a new iteration. The artifact becomes a living loop.

Planned feature

Export with metadata thread

Each export carries embedded identifiers mapping cells/slides back to the original model assumptions.

Stakeholder edits externally

CFO changes WACC from 10% to 9.2% in the exported .xlsx. VP reorders three slides in the .pptx.

Re-ingest and diff

Drop the edited file back into iterFact. The engine parses what changed without breaking the model structure.

New iteration created

v2 is stamped. Every view, every chart, every downstream export reflects the reconciled assumptions.