COMPARE METHODS
Goal Seek vs Solver: choose the problem before the method
Use a one-input Goal Seek when the financial question leaves one assumption free. A solver-style problem is needed when several assumptions may change together or derived outputs must satisfy additional constraints. Adding variables increases the feasible choices; it does not automatically produce a better economic decision.
| Question | One-input Goal Seek | Multi-variable / constrained problem |
|---|---|---|
| Typical task | Find an entry multiple for a target IRR | Find pricing and financing assumptions that meet a return and leverage condition |
| Changing inputs | One | Several selected inputs |
| Restrictions | An allowed range for the input | Input bounds plus constraints on registered outputs |
| Review | Target gap and changed assumption | Target gap, every constraint and the full changes |
| Meaning of success | A candidate meets the configured target tolerance | A tested candidate meets the configured goal and constraints |
On this page
The spreadsheet terminology describes a real difference
Microsoft documents Excel Goal Seek as changing one input to reach a formula result. Its Solver supports multiple changing cells, an objective and constraints. That is a useful distinction in problem structure; it is not a claim that BalanceCheat uses the same algorithms, menus or every constraint type.
BalanceCheat exposes these problems through the Reverse Layer and registered financial inputs and outputs. The one-input workflow is described in Financial Goal Seek. This page helps decide when that narrower setup is insufficient.
A bound and a financial constraint do different jobs
An input bound says what a variable may be. A derived constraint says what the evaluated model must satisfy. A permitted debt-leverage input of 0×–5× does not by itself guarantee that the resulting financing meets a 4× leverage limit.
For the tested synthetic LBO configuration, the derived leverage metric is entry gross debt divided by the entry metric. It is not Net Debt / EBITDA. Keep the actual registered metric name and basis when specifying a limit.
| Component | Configured condition | Role |
|---|---|---|
| Goal | LBO IRR ≥ 25% | Desired result |
| Variable 1 | Entry Multiple: 5×–9× | Permitted pricing range |
| Variable 2 | Debt leverage input: 0×–5× | Permitted funding range |
| Derived constraint | LBO entry gross leverage ≤ 4× | Restricts evaluated financial outcomes |
Read feasibility before interpreting the return
The native test finds a candidate with IRR 25.0000% and entry gross leverage 4.0000×. The leverage constraint is binding within its configured numerical tolerance of 0.00000001×. This verifies that the condition participates in the solve; it does not prove that the resulting capital structure is optimal.
A result can meet the return but fail a constraint, or satisfy all constraints while missing the return. Read both statuses. If several candidates satisfy the request, the selected result depends on the search settings and preference for small changes relative to the configured ranges.
What the current native Layer supports
Reverse uses registered eligible inputs, one primary target, bounded variables and additional registered scalar constraints. Target conditions include equality, thresholds and ranges; optimization modes seek to improve a target. Current limits are twelve variables and ten additional constraints. These are product limits, not a recommendation to use the maximum.
Source ownership still matters. Writable assumptions can be applied after review; analysis-only intermediate shocks cannot be silently committed. Results become stale after a model change. No result promises global optimality, a proof of infeasibility or an irreducible conflicting-constraint certificate.
Choose the smallest financially complete problem
Keep one variable when the other assumptions are decisions already made. Add variables when they genuinely remain negotiable. Add a derived constraint when a financial limit must hold after recalculation. Avoid broadening bounds merely to make the status turn successful.
For controls and precise limits, use multiple variables, bounds and constraints and result status. The Analyze overview shows when a forward sensitivity test is the more useful question.
Sources and conventions
Synthetic examples. Model-based results were reconciled with the current BalanceCheat calculations; assumptions, units and boundaries are specified in each example.