How these numbers are calculated

All seven tools read the same inputs and call the same calculation engine. The engine is deterministic: the same inputs always produce the same output, on any device, on any day. It contains no randomness, no simulation and no language model — a language model writes none of the arithmetic, at build time or while you use the site.

Plain-English explanation sits on top of the numbers, never in place of them. Where the tools describe what a result means, they are describing a figure the engine already computed.

Your balance is compounded forward to the first day of college, with contributions added along the way. Three conventions decide the answer, and all three are easy to get wrong quietly:

• Contributions land at the start of each month — an annuity-due — and stop the day education begins. A contribution made in January earns for that whole month.
• The monthly rate is the twelfth root of the annual rate. A stated 6% therefore earns exactly 6% over a year. Dividing by twelve instead — the common shortcut — would quietly return 6.17% on a 6% assumption, and would flatter every projection on the site.
• Each college bill is paid at the start of its year, and whatever remains keeps earning through the years that follow. A four-year drawdown is not four equal withdrawals from a frozen pot.

College is assumed to begin at age 18 unless you say otherwise.

Every default is visible and adjustable on the page. These are the starting points:

• 6% annual return. A middle-of-the-road assumption for a diversified, age-appropriate portfolio. Results are also shown 2 points either side of it, because a single number implies a confidence nobody has.
• 4% college cost inflation. College Board’s own 54-year series puts real growth at roughly 1.5–2% a year; 4% nominal is about 1–2% real, which is mildly conservative rather than alarmist. Recent decades have been flatter than the long series.
• 4 years of college. Adjustable, and worth adjusting — completion in four years is not the norm everywhere.

Cost figures are College Board’s average estimated full-time undergraduate budgets for the 2025–26 academic year, rounded to the nearest thousand. A budget covers tuition and fees, housing and food, books, transport and other expenses — not tuition alone.

• Public, in-state: $31,000 a year
• Public, out-of-state: $51,000 a year
• Private: $65,000 a year

Averages hide an enormous spread — published in-state tuition alone runs from roughly $6,360 to $18,090 depending on the state — which is why every tool lets you enter your own figure instead. Source: College Board, Trends in College Pricing, last verified 22 August 2026.

Some tools report a coverage ratio and some a funding ratio. They are different questions and they meet only at 1.

Coverage is the share of the actual bills your money pays, capped at 100% — you cannot cover more than all of it. Funding is your day-one balance against your day-one need, and it is uncapped, so it can read 140%. Neither is wrong, and one is not a broken version of the other.

Nothing rule-derived on this site is written from memory. Tax treatment, contribution limits and state rules are transcribed from a primary source, with the URL and the date it was last checked recorded next to the figure.

Where a rule cannot be sourced, the tool says so on the page and declines to produce a number, rather than guessing and presenting the guess in the same typeface as a fact. You will see this most often in the leftover-money tool, where state treatment varies sharply and several states are not clearly documented anywhere authoritative.

Stated plainly, because a limitation you discover later reads as a defect:

• State income tax benefits are not modelled. Many states offer a deduction or credit for 529 contributions. Getting this right for 50 states requires per-state sourcing that is not finished, so it is absent rather than approximated. Where it would matter, your real position is better than the tools show.
• Financial aid treatment is not modelled. A 529 owned by a parent affects aid calculations differently from one owned by a grandparent. The tools do not attempt this.
• The brokerage comparison is simplified. On the 529-versus-brokerage tool, tax on the brokerage side is computed as gain over principal, taxed once. That omits the annual drag of taxable dividends — which understates the 529’s advantage rather than overstating it. The simplification is labelled on the page itself.
• Returns are assumptions, not forecasts. Nothing here predicts a market. You supply a rate; the engine applies it consistently.

The engine is covered by an automated test suite that runs on every change, and its results are verified against an independent implementation written separately from the engine itself. That practice has caught real errors before release — including a compounding convention that returned 6.08% on a 6% assumption, and a tax split that quietly flattered the taxable account.

If you work through a projection by hand and get a different answer, that is worth telling us about. Please get in touch.