How Finance Outcomes calculates.
The numbers come from deterministic TypeScript finance engines—not AI-generated mathematics. Each calculator page publishes its formula, assumptions, limitations, and known-answer examples.
Calculation path
- You enter a scenario in a labeled form.
- The browser validates the input and passes it to an independent finance engine.
- The engine returns full-precision numerical results.
- The interface rounds only for display and explains the result.
Shared conventions
- Entered annual rates are assumptions and remain constant unless a page says otherwise.
- Money inputs within a calculator must use the same currency; no foreign-exchange conversion occurs.
- No calculator fetches live prices, rates, or account data.
- Taxes, fees, inflation, timing, and cash flows are included only when the page explicitly models them.
Deterministic driver sensitivity
Controlled reruns
For each eligible driver, the current valid scenario is recalculated after a controlled 10% change to that one input while other inputs stay fixed. Zero values use a documented input-specific fallback step. Whole-year fields remain whole numbers, bounds are respected, and invalid perturbations are skipped.
Normalized comparison
The absolute percentage change in the primary modeled output is divided by the proportional input change. A score of 0.75 or more is High, 0.25 to below 0.75 is Moderate, and below 0.25 is Low. Bars visualize the score and are capped only for display.
No AI model ranks the drivers. The reusable sensitivity module reruns the same deterministic finance engines used by the calculators. The labels describe local scenario sensitivity, not probability, certainty, risk tolerance, tax eligibility, or a recommendation.
ROC lookback is shown as context rather than given a fake numerical score: changing the lookback number alone cannot change ROC unless the corresponding historical reference price also changes.
Known-answer and boundary testing
Automated tests compare engine output with independently known or algebraically derived results. They also verify ordering relationships, zero-value branches, finite-number checks, supported bounds, and invalid-input rejection. Tests run before a release is accepted.
Interface formatting never feeds back into the calculation, so displayed rounding cannot compound through later steps.
Read the method beside each tool
Investment Drawdown
Formula, conventions, assumptions, limitations, and known-answer checks.
Drawdown Recovery
Formula, conventions, assumptions, limitations, and known-answer checks.
CAGR
Formula, conventions, assumptions, limitations, and known-answer checks.
Average Cost / Cost Basis
Formula, conventions, assumptions, limitations, and known-answer checks.
Dividend Income
Formula, conventions, assumptions, limitations, and known-answer checks.
Tax-Loss Harvesting
Formula, conventions, assumptions, limitations, and known-answer checks.
Investment Fee Impact
Formula, conventions, assumptions, limitations, and known-answer checks.
FIRE Number
Formula, conventions, assumptions, limitations, and known-answer checks.
Coast FIRE
Formula, conventions, assumptions, limitations, and known-answer checks.
Rate of Change (ROC)
Formula, conventions, assumptions, limitations, and known-answer checks.
Mortgage vs Invest
Formula, conventions, assumptions, limitations, and known-answer checks.
Educational scenarios: results are modeled estimates, not forecasts or personalized financial, investment, legal, or tax advice.