How the numbers work

RunwayFI is a reactive digital twin of your finances, not a calculator. Every fact you enter flows through one computation graph: change an input and every downstream figure — net worth, taxes, your FIRE age, days of freedom — recomputes on its own. There is no “recalculate” button and no manual refresh, because there is nothing to refresh; the chart is always showing the consequence of your current numbers.

The tax math runs under real, current-year law. Brackets, contribution limits, Social Security figures, and Federal Poverty Level values are not hardcoded in the app — they are data, mirrored from the IRS, SSA, and HHS through the open-source PolicyEngine model, re-pulled on a schedule with drift detection before any change reaches the engine. The details of how future years are projected (published values, a CBO forecast horizon, then a long-term estimate) are in the Tax Projections Methodology.

One rule holds everywhere: the engine does all the math. The chart, the tiles, and PlanPulse (the AI companion) only present engine-computed numbers — they never invent their own. So a figure means the same thing on every surface it appears.

The one error that would make a planning tool worthless is telling you that you can retire when you cannot. RunwayFI is built to fail in the safe direction: when there is a choice, it shows the more conservative figure.

• It rounds down, never up. Your spendable cash (Safe-to-Spend on Track) is available minus committed bills minus your buffer, floored to the whole dollar and clamped at $0. Rounding a Safe-to-Spend figure up would overstate the cash you have; a shortfall shows $0, never a negative that could look like room.
• Today is a strict boundary. A future-dated transaction is invisible to today’s Safe-to-Spend, so income you have not actually received yet can never inflate what the app says you can spend now.
• A missing fact lowers confidence — it never guesses high. If a required input is absent, the figure is blocked and the tile asks for it (“Needs income”, “Add an account”) instead of filling the gap with an optimistic guess. If some of your cash cannot be verified through today, the number drops from “Clean” to “Partial” rather than quietly counting cash it is unsure of.

When a number is an estimate rather than a settled figure, RunwayFI marks it with ≈ and tints it amber. The current year is exact; a projected FIRE age or a future balance carries the ≈ because it depends on assumptions about years that have not happened. (The glyph is always ≈, never a tilde — at small sizes a tilde reads like a minus sign, and that is a trust defect on a money surface.)

Behind the symbol is a four-level Data Assurance Level (DAL) that says how solid a figure is:

• DAL 1Blocked — missing required inputs
• DAL 2Estimate — some inputs are approximated
• DAL 3Partial — all required inputs present
• DAL 4Clean — all inputs verified

Confidence only ever propagates downward. An estimate feeding a calculation can never produce a “Clean” result — a figure is only ever as trustworthy as its weakest input. That is why adding one real fact can visibly sharpen a whole column of downstream numbers.

Numbers on the chart and in the “Show your math” view carry a colored provenance badge so their source is never a mystery. There are four:

• Input — A number you entered — a balance, your income, your monthly spending.
• Computed — A value the engine derived from your inputs and the rules — never stored, always recomputed.
• IRS rule — A cited constant from the law — a bracket, a contribution limit, a poverty-line figure.
• Assumption — A planning assumption or a forward projection — a return rate, an inflation rate, a future year.

For tax figures there is a second, source-of-law axis. Each projected year is labeled by how solid its underlying rules are — IRS published (values the IRS, SSA, or HHS have officially announced), CBO forecast (inflation-adjusted within the CBO’s ten-year horizon), or Long-term estimate (rules held constant beyond that horizon). If the rule lineage for a year cannot be verified at all, an amber banner says so plainly — “Tax rule provenance unavailable” — rather than showing a confident number on shaky ground. The Tax Projections Methodology walks through each tier.

RunwayFI explains the consequence of a decision in your numbers; it does not tell you what to do. That is a deliberate line, and it is enforced in the product, not just promised here. The AI companion is held to a fixed voice — it never says “you should”, “I recommend”, or “I advise”, and always frames things as “the data shows”, “you might explore”, or “based on your numbers”. Those directive phrases are blocked at runtime, so a slip cannot reach you.

The same discipline keeps RunwayFI on the safe side of financial regulation: it plans around tax brackets, ACA cliffs, IRMAA tiers, and account types — it never names a specific security, fund, ETF, or brokerage, and it is not a registered investment adviser. Every page that shows computed values carries this note:

• No return promises. Nothing here is guaranteed. Projections are a hypothetical illustration of math applied to your inputs and each year’s rules — not a prediction of your actual future, and never a promised outcome.
• No personalized investment advice. RunwayFI does not recommend securities, funds, or portfolios, and it is not a substitute for a professional.
• It is not a budgeting app, a chatbot, or a calculator. It is a decision-support tool that explains what a choice does to your runway, with the math showing and the uncertainty visible.

New here? The Getting started guide walks you from sign-up to your first honest runway number.