Data Cocoon LLC — B4_UR_IZ

Somewhere between Bloomberg terminals, Excel macros, and six-figure enterprise licences, we lost something obvious: the ability to actually see a company.

I don’t mean a chart. I don’t mean a pivot table or a dashboard with twelve KPI widgets competing for attention. I mean the architecture of the thing—the way revenue cascades into cost structures, how balance sheet positions flow through cash flow statements, where value concentrates and where risk hides. The structural skeleton that every analyst, auditor, and CFO carries in their head but has never once been able to look at.

Until now. A system called VTM—Visual Tree Model—does something that no tool in the financial technology ecosystem currently does: it renders a company’s complete financial structure as a three-dimensional model you can navigate in real time, run valuations against, stress-test with Monte Carlo simulations, decompose through DuPont and ROI analysis in three dimensions, project forward with statistical forecasting, and interrogate node by node, all inside a single self-contained HTML file that loads in your browser in under three seconds.

The Problem Nobody Solved

Financial data is fundamentally hierarchical. Revenue breaks into segments. Segments break into product lines. Product lines carry cost structures that roll up into gross margins. Balance sheets nest current assets inside total assets inside total equity plus liabilities. Cash flow statements reconcile net income through operating, investing, and financing activities. SEC Note Schedules—warranty reserves, deferred revenue, debt structures, lease obligations, equity compensation, digital assets—each carry their own sub-hierarchies.

Every tool in the current ecosystem treats this data as flat. Excel shows rows and columns. Bloomberg shows panels. Tableau shows charts. Power BI shows dashboards. Even the most sophisticated financial modelling platforms—Capital IQ, FactSet, Refinitiv—present hierarchical data as nested tables that collapse and expand. None of them show you the shape.

VTM does. It maps every financial line item onto a polar coordinate system—concentric rings radiating outward by depth level, angular partitions by sibling position, height by value. The result is a structure that looks less like a spreadsheet and more like the cross-section of a living organism. Because, in a real sense, that’s what it is.

What It Actually Does

The current production build of VTM—a single HTML file of approximately 39,000 lines—contains the following capabilities, all running client-side with zero server dependencies:

The 3D Studio

A WebGL-powered Three.js environment that renders the complete financial hierarchy of any company as a navigable 3D structure. For the Tesla proof of concept, this means 234 nodes across Income Statement, Balance Sheet, Cash Flow, Equity Changes, 17 SEC Note Schedules, 3 operating segments, and 15 financial KPIs, all positioned mathematically by their structural relationships. Nodes are colour-coded by financial statement type and sized proportionally to value. The user can orbit, zoom, drill into sectors, isolate subtrees, and scrub forward and backward through 26 periods of historical data—or 36 periods when forecast projections are active.

A ring-based isolation system allows the user to selectively show or hide nodes at each depth level. Ring 0 isolates the root parent node independently—the apex of the financial tree. Rings 1 through 5 correspond to successively deeper structural layers, from major financial statement categories down to individual line items. Each ring can be toggled independently across all ten 3D model types, giving complete control over structural visibility.

The 2D Dashboard

A tabular analytical view of the same data, toggleable in one click. Every row is linked to its 3D counterpart. Click a node in the table; the 3D model highlights it. The dashboard includes full multi-column sorting, filtering by financial statement type, depth-level drill-down, share price input integration for market-cap-dependent metrics, and a five-schedule ledger system: Base Data, Journal Entry Deltas, Projections, What-If Percentage Adjustments, and a Computed Result schedule that combines them all with automatic balance verification. This is a genuine double-entry system running in a browser.

The Valuation Engine

Eight distinct valuation methodologies run against the live dataset: DCF (Discounted Cash Flow) with adjustable WACC, terminal growth rate, and projection period. Appraisal-Based valuation using asset and liquidation approaches. Sum-of-Parts decomposition by operating segment with per-segment multiples. Returns Analysis benchmarking ROE, ROA, and capital efficiency. Comparable Multiples using P/E, P/B, EV/EBITDA, P/S, and P/FCF against peer inputs. DuPont Decomposition breaking ROE into its three-factor and five-factor components. A 10,000-run Monte Carlo simulation that randomizes WACC, terminal growth, and projected cash flows across normal distributions to produce fair value confidence intervals. And a unified export panel that compiles everything into downloadable reports.

DuPont Decomposition in Three Dimensions

This is new, and it changes the way you think about return on equity.

Traditional DuPont analysis presents ROE as a formula: Profit Margin × Asset Turnover × Equity Multiplier. You see three numbers. Perhaps a chart. You know intellectually that these three ratios interact, that leverage amplifies both returns and risk, that margin compression can be offset by efficiency gains—but you have never seen it.

VTM’s DuPont 3D module constructs a hierarchical tree from the DuPont decomposition and renders it as a navigable three-dimensional model. Depth 1 is the ROE apex—a single node representing the consolidated return on equity across all periods. Depth 2 fans out into the three DuPont factors: Net Profit Margin, Asset Turnover, and Equity Multiplier. Depth 3 exposes the raw components beneath each factor.

The result is startling. You can see, spatially, how a company’s return profile has evolved over a decade. A widening margin cone tells you profitability is improving. A flattening turnover band tells you asset efficiency has stalled. A leveraging spike tells you the company took on debt to amplify returns. These are patterns that a table of numbers takes minutes to decode. In the 3D DuPont model, they are visible in a single glance.

The DuPont 3D model loads from a dedicated button and coexists with the standard holistic model. A RESTORE button returns the user to the full company view at any time, with complete state preservation across all five schedules—base data, journal entries, projections, what-if adjustments, and computed results are all deep-cloned before the DuPont model loads, and fully restored when the user returns. No data is lost. No state is corrupted.

ROI Decomposition: From Diagnosis to Prescription

If DuPont tells you what happened to equity returns, ROI tells you what to do about it.

The ROI Decomposition module computes eleven return-on-investment metrics per period and renders them both as an analytical dashboard and as a three-dimensional model. The centrepiece is ROIC: Return on Invested Capital, decomposed as NOPAT Margin multiplied by Capital Efficiency. This is the metric that separates companies creating value from those destroying it.

The module computes: ROIC, ROA, ROE, Cash ROIC, Gross ROA, Operating Margin, EBITDA Margin, Asset Turnover, Capital Efficiency, and Debt-to-Capital ratio. The Value Creation Analysis computes the spread between ROIC and WACC and presents it as a direct measure of economic value creation. A positive spread means the company earns more than its cost of capital; a negative spread means it is destroying value with every dollar invested.

The 3D ROI model mirrors the DuPont structure: a depth-1 apex node representing consolidated ROI, depth-2 nodes for each major metric, and depth-3 nodes exposing the numerator-denominator components. The colour mapping uses a distinct green hue (145°) to differentiate it visually from the DuPont model’s blue-purple palette. Like DuPont, the ROI model includes full backup-restore functionality with deep cloning across all schedules.

The Observatory

An analytical layer that answers structural questions: which nodes are growing fastest? Where is value concentrating? What is the CAGR across every branch of the hierarchy? How does volatility distribute across the tree? The Observatory uses multi-pass bottom-up aggregation to cascade leaf-node values through container nodes, then computes growth rates, composition percentages, and trend analysis across the full hierarchy.

The Statistical Projection Engine

VTM’s FORECAST system generates forward projections for every node in the tree, selecting methodology per node based on the statistical profile of each time series. Five projection methodologies are available: HIGH_GROWTH_DECAY for nodes with recent CAGR above 15%. STABLE_GROWTH for consistent positive trajectories. RECOVERY_TRAJECTORY for recent downturns. MEAN_REVERSION for high-volatility nodes. And CAGR_DAMPENED as the default.

A separate PROJECT control offers simpler growth projections—Low (1–2%), Medium (3–5%), or High (6–10%)—using three regression methodologies. The PROJECT and FORECAST systems coexist cleanly.

Ratio-Aware Forecasting for Valuation Models

VTM solves the problem of flat ratio projections by projecting the ratios themselves forward, classified by type. Profitability ratios use mean-reversion toward the midpoint between recent average and historical peak. Efficiency ratios project toward the upper quartile of historical range. Leverage ratios mean-revert toward the historical average with capped annual movement.

Forecast Data Synchronisation

VTM addresses stale valuation data with a three-point synchronisation system ensuring that DuPont, ROI, and all valuation models always reflect the most current Sch5 data, regardless of the order in which the user operates.

The Five-Schedule Backup Architecture

VTM’s backup system deep-clones all five schedules—Sch1 through Sch5—along with all state variables. The backup is created only once: the first model load captures the original state. Restoration is complete and instantaneous. The user returns to exactly the state they were in before loading any decomposition model.

And Everything Else

Voice narration with natural speech synthesis. Annotation bookmarks with persistent storage. A blockchain-stamped audit trail for data integrity verification. Multi-screen synchronisation via BroadcastChannel. VR-ready rendering. A SQL query engine for ad hoc data interrogation. A General Ledger import module. Scenario comparison. CSV/JSON import-export. A complete in-app user manual. All of this inside a file you could email as an attachment.

What Makes This Different

It is not the individual features that matter. What does not exist—anywhere, in any product at any price point—is the following combination:

Zero Infrastructure	No server, no database, no API key, no subscription, no installation. Open a file. A regulator in Lagos, a student in Jakarta, and a fund manager in Zurich all have equal access to the same analytical power.
Structural Fidelity	The 3D model preserves the actual hierarchical relationships defined by GAAP and IFRS reporting structures. You see the tree as the tree.
Full-Stack in a Single File	Visualisation, valuation, DuPont decomposition, ROI analysis, statistical forecasting, stress-testing, audit trail, and reporting in one self-contained deliverable. The file IS the application.
Dataset-Agnostic	Load any hierarchical financial dataset—any company, any structure, any reporting standard. The system auto-detects node types and calibrates accordingly.
Bidirectional Navigation	Click a node in 3D and the table highlights. Click a row in the table and the 3D model focuses. Every analytical finding is spatially addressable.
Auditable by Design	Five-schedule ledger architecture with balance verification. Blockchain-stamped audit trail. Deep-cloned backup and restore across all schedules. Every adjustment is traceable.
Analytical Depth	DuPont and ROI decomposition rendered as interactive 3D models. Ratio-aware forecasting. Value creation analysis measuring ROIC against cost of capital. Capabilities that dedicated platforms charge six figures for.

Why It Matters

The financial analysis industry runs on tools designed in the 1980s and refined incrementally since. The spreadsheet is forty years old. The terminal is thirty. The cloud dashboard is fifteen. None of them were designed for the data volumes, structural complexity, or speed requirements of modern financial reporting.

VTM is not a replacement for Bloomberg or Excel. It is a genuinely new category—a spatial financial analysis system that treats the structure of financial data as a first-class object. It is, to the best of my knowledge and research, the only system in existence that renders a complete SEC filing structure as an interactive, navigable, analysable three-dimensional model.

It is also the only system that renders DuPont decomposition and ROI analysis as three-dimensional navigable structures. Seeing ROE as a spatial object—where margin is a dimension, turnover is a dimension, and leverage is a dimension—changes the way you think about return on equity.

It was built on Tesla’s public filings because Tesla is one of the most structurally complex public companies. If VTM works on Tesla, it works on anything.

The Invitation

I am not a software engineer. I am a finance professional who needed to see something that no existing tool could show me. So I built it—with the help of AI, with an obsessive attention to getting the numbers right, and with the conviction that the way we visualise financial data has been fundamentally inadequate for decades.

The system is real. It works. You can load any company’s financial data, and within seconds you are flying through its financial architecture in three dimensions, running valuations, decomposing returns through DuPont and ROI analysis, stress-testing assumptions, projecting forward with statistically grounded forecasts, and identifying exceptions that would take an analyst days to find in a spreadsheet.

One file. No server. No subscription. No dependencies.

Thirty-nine thousand lines of code that fit in an email attachment and do the work of a six-figure analytics platform.

The spreadsheet had a good run. It is time to see the shape of the data.