The spec, not a forecast

Methodology and assumptions

Updated . The pages and the unit tests call these functions.

Clew plans in today’s dollars. You type spending in today’s prices. The calculator removes inflation from the return instead of inflating every future bill. Changing a field recalculates on your device. The first paint of each example is the same code, rendered into the HTML so you can read a number before JavaScript runs.

Real return

Real return = (1 + nominal return) ÷ (1 + inflation) − 1. It is not nominal minus inflation, though the two are close when both rates are small. The default example is 7% nominal and 2.5% inflation:

(1 + 7%) ÷ (1 + 2.5%) − 1 = 4.39%

A negative real return is allowed. In that case the Coast FIRE number is larger than the FIRE number, because the portfolio is expected to lose purchasing power and you would need more today to finish with enough.

FIRE number

FIRE number = annual spending ÷ withdrawal rate.

$60,000 ÷ 4% = $1,500,000 (25× spending)

The 4% default is the familiar rule of thumb from William Bengen’s research and the Trinity study: historical US stock-and-bond mixes over roughly 30-year retirements. It is not a promise, and it is a weaker fit for a 40- or 50-year retirement or for spending you cannot cut. The field accepts 0.5% to 12% so you can see the multiple move. Clew does not pick a “correct” rate.

Coast FIRE number

Coast FIRE number = FIRE number ÷ (1 + real return) ^ years until retirement.

$1,500,000 ÷ (1 + 4.39%)^33 = $363,341

Years are (retirement age − current age), measured in months and converted back so a fractional age still discounts cleanly. If retirement age equals current age, the coast number is the FIRE number. You have hit Coast FIRE when invested assets are at least that number, within fifty cents of rounding tolerance.

Why zero contributions never close a gap

Let P be the portfolio and C the coast number today. With no contributions, both grow by the real return. After t years the portfolio is P × (1 + r)^t and the coast target is C × (1 + r)^t. The ratio P / C is constant. Tests lock this property. Contributions are the term that breaks it.

Monthly contributions

The annual contribution is split into twelve equal amounts in today’s dollars. Each month the balance is multiplied by the monthly rate that compounds to the annual real return, (1 + r)^(1/12) − 1, and then the monthly contribution is added. The coast target is checked after that contribution. The reported age is your current age plus the number of months until the first passing check. If you already pass, the age is your current age, before any new contribution.

The chart samples that path every three months, plus the crossing month and the final month. The table shows each year. “Save $X a year to reach FIRE by retirement” is the level payment, continued every month until retirement, that lands on the FIRE number. It is not the same as “pay this until you hit coast and then stop,” which would be a shorter, larger contribution. Hitting coast means you are allowed to stop. The suggested savings rate assumes you do not.

FIRE, lean, and fat projections

The FIRE number is constant in today’s dollars, so unlike the coast target it does not rise with age. The projection steps month by month until the portfolio reaches it, up to 50 years or age 90, whichever comes first. The chart then shows the path through the hitting point plus two years, and at least 15 years, so a quick success does not collapse the picture. Lean and fat pages use this same projection. Only the default spending, age, and contribution change. Those defaults are labels, documented on each page, not definitions.

Barista FIRE

Portfolio spending = max(0, annual spending − part-time income). Barista FIRE number = portfolio spending ÷ withdrawal rate. Full FIRE number = annual spending ÷ withdrawal rate. If income covers spending, the barista number is $0 and you have “hit” it only in the sense that the portfolio is not being asked to pay. The income is assumed to continue, in today’s dollars, for the whole drawdown. The calculator does not model a job that ends on a date.

What is left out

  • Taxes on contributions, dividends, and withdrawals. Put after-tax spending in the spending field if you want the portfolio to cover the tax too.
  • Fund fees and advice fees. Use a nominal return that is already net of the expense ratio you expect.
  • Social Security, pensions, annuities, and a partner’s income. Subtract them from spending yourself if you want them in the plan.
  • Housing you will not sell. It is not part of “invested assets” unless you decide it is.
  • Sequence of returns, crashes, and inflation spikes. One real return every year is a teaching model. It is not a probability.
  • Healthcare before Medicare, and the shape of spending as you age. Spending is flat in today’s dollars for the whole horizon.
  • Employer matches, contribution limits, and required minimum distributions.

The unit tests live next to the functions and cover the closed form, the constant-ratio property, the contribution that lands on a future balance, barista income, and the shared URL fields. If a result on the page and a sentence on this page ever disagree, the function is the source of truth and this page should be updated to match it.