Stochastic: uncertainty through the model

Methods and settings

Build a simulation deliberately

First validate the deterministic model. Select a target, then configure distributions on eligible sources using the σ controls or the Layer’s input selection. Inspect the input’s identity, current value, unit and ownership. Add only the inputs whose uncertainty belongs in the question. Choose sampling settings, optionally set a seed, and run.

The current UI supports up to 100 saved source configurations per model and 20 active inputs in a run. These limits are not a recommendation to use the maximum: additional uncertain assumptions can make interpretation harder.

Ask a conditional risk question

For a supported liquidity target, define uncertainty on eligible cash or operating assumptions and examine the probability of falling below a stated threshold. For an eligible valuation target, inspect the distribution of Equity Value under the selected assumptions. The resulting probabilities describe your model and configured distributions, not independently verified market probabilities. Domain-specific eligibility still applies.

Read validity before percentiles

Review valid and invalid run counts before interpreting the histogram, CDF, statistics, thresholds or input ranking. An invalid result is excluded from target statistics, which can materially change the population being summarized. Cancelled runs show only their completed subset. Model edits make existing results stale; re-run on the revised model. Configurations persist, but completed sample arrays and charts are not a saved simulation library.

Understand the uncertainty boundary

Inputs are sampled independently; there is no user-defined correlation matrix. Modeled dependencies still recalculate, so a sampled revenue source can affect linked costs and cash. That economic linkage is different from correlated external shocks. No automatic distribution fitting, time-series forecasting or causal statistical inference is provided. Use dependencies and correlation before combining related assumptions.

Public models do not store this analysis session

Stochastic distributions, run configurations, samples and results are not persisted in public model snapshots. Publication preserves operator inputs and deterministic operator outputs. After Fork, run this analysis again locally against the restored editable inputs. A reviewed change explicitly applied to an underlying operator input can be published as that input; the analysis session that produced it is still excluded.