Model Library: local models, public models and Fork

Methods and settings

Two libraries, three kinds of model state

The public Model Library is a repository of published BalanceCheat financial models. Readers can open, inspect and Fork a model without an account. The Library offers search and filters; an optional ticker helps identify an analysis but does not verify the issuer or fetch a share price. Its EN/DE controls change the interface language; the theme control changes presentation. Publisher-written text is not automatically translated.

Your local browser library remains the editable workspace. Use a separate model for another company and scenarios for supported alternative assumptions within a model. Local duplication makes an independent working copy; it is not a live connection or a new public publication.

Know where the data is retained

Local models are stored in the current browser profile. They remain local unless you explicitly publish them or send content through a separately authorized export or AI connection. Clearing site data can remove local entries; use .balancecheat exports for portable backups.

Publishing creates a separate immutable public snapshot of selected operators, scenarios and required dependencies. The public URL can be shared, found through the Library and indexed subject to search-engine behavior. Later local edits do not change that version. Publishing and Forking need no user account in V1; this does not create private cloud storage or an account-owned editing space.

Delete deliberately

Rename or delete a local entry through model management. Deleting the last local entry creates a blank working model; downloaded files and published versions remain separate. V1 has no publisher edit, delete or unpublish control for public snapshots. An administrator can remove or restrict public content under the existing moderation process. Do not publish information that must later remain under your personal deletion control.

Share a financial model by URL

Use Publish and share the resulting public URL. Unlike sending a spreadsheet file, the link opens a structured read-only financial view: readers can inspect inputs, outputs and supporting schedules, then Fork the model to work on its assumptions. The shared snapshot does not silently change when you edit your local model.

This is public sharing, not simultaneous editing, comments, social features or private cloud collaboration. A local file can still be useful for a private transfer through a channel you choose; a public URL has a different disclosure boundary.

Review inputs, outputs and disclosures

A public page shows the title, optional ticker, description, operator scope, scenarios, horizon, currency/units, publication date and direct parent when available. Compact financial tables present key outputs; expandable schedules and included thesis, assumptions and notes provide context. Advanced details expose native inputs and the exact snapshot. Raw stored amounts can use full currency units while compact tables use the displayed scale; read the unit labels.

The publisher's local engine produces the calculations. The server validates structure and publication rules; it does not independently verify financial correctness, issuer identity or investment quality. Reading the tables does not require JavaScript; interactive Fork and Report do. Use Report for the available abuse/privacy reasons, not to edit a publication or guarantee its removal. The informational and educational disclosure applies to public stock analyses.

Fork and continue another model locally

Choose Fork on a public page or Library card. BalanceCheat reads the published snapshot, verifies it against its retained calculation runtime and adds an editable local model. Forking performs no publication/database write: it downloads and restores content in your browser. If all five local slots are occupied, export and remove an unneeded local entry first; a failed Fork must not replace existing work.

The new copy contains the selected published operator inputs, not the publisher's unrelated local models or omitted operators. Local edits are independent. You can run Analyze again, export the copy or later publish a new snapshot. Fork is not an edit to the original URL.

Direct-parent provenance

A Fork retains the public model's identity and snapshot hash as its direct parent. This origin survives supported local duplication and model-file export/import. Publishing that copy can show a link back to the parent; Forking the new publication makes that new publication the next direct parent. This is direct-parent lineage, not a full version graph, attribution of authorship or proof of review. If a parent is unavailable, the page identifies that condition rather than inventing its content.

Public operators, local Analyze sessions

Public snapshots persist selected operator inputs and deterministic operator outputs. They do not persist Inspector pins/follow state, Key Drivers rankings, Sensitivity setup or grids, Tornado configurations/rankings, Stochastic distributions/runs/results, or Reverse / Goal Seek solver state and candidates. Expanded panels and temporary selections are not public model content.

Underlying durable operator assumptions needed for calculation remain included. After Fork, run Analyze again against those editable inputs. Local model files may retain supported analysis configurations; that does not mean those configurations are published. A public valuation range is an operator output, not a saved interactive analysis session.