Every number in FIRE Journey comes from one of three engines that share the same model of your plan: your incomes, expenses, account balances, and assumptions. This page explains what each engine does, where the data comes from, and where the model deliberately simplifies.
The deterministic projection walks your plan forward one path at a time using your assumed average returns. The portfolio compounds monthly; taxes are computed once per year. This engine produces the projection chart, the year-by-year ledger, and your FIRE date.
The historical backtest replays your retirement against every rolling cycle in the historical record: retiring into 1928, 1929, 1930, and so on through the last start year that fits a full retirement. Each cycle uses the real sequence of stock, bond, cash, and inflation numbers from those years. Success means the portfolio never depleted.
The Monte Carlo simulation runs 10,000 seeded simulations. Instead of drawing returns from a bell curve, it bootstraps blocks of five consecutive historical years, so crashes and recoveries keep their real shape and the correlation between stocks, bonds, and inflation is preserved. Each series is shifted so its long-run average matches your plan's return assumptions.
The three engines answer different questions, so their numbers can disagree. The projection assumes you earn the average every year; the backtest and Monte Carlo test what happens when returns arrive in a bad order. A plan can look fine on the average path and still fail in a third of simulations because of sequence-of-returns risk. That disagreement is the point.
The backtest and Monte Carlo share one annual series covering 1928 through 2025. Stock returns are S&P 500 official total returns (S&P Dow Jones Indices). Bond and cash returns are the 10-year Treasury bond and 3-month Treasury bill annual returns from Aswath Damodaran's dataset at NYU Stern. Inflation is BLS CPI-U, December over December. We update the series each January when all four sources publish.
For US households, projections run a pro-forma federal return for every year: ordinary brackets, the standard deduction, long-term capital gains stacked on ordinary income with their own breakpoints, Social Security taxed through the provisional-income formula, the net investment income tax, the child tax credit, and FICA during working years. All constants come from the official 2026 tables and are indexed forward with your assumed inflation; the Social Security thresholds stay frozen, as they are in law. State and local tax is a flat rate on AGI minus taxable Social Security, since most states exempt it.
Deliberate simplifications: standard deduction only (no itemizing), no married-filing-separately, no AMT, and a flat state rate rather than state brackets. Households outside the US get an explicitly simplified model instead of silently inheriting US rules: one effective tax rate you control, with no RMDs, early withdrawal penalties, or state tax.
Retirement spending is sourced from your accounts in tax order: taxable first (only the gains are income), then traditional (fills the low brackets), then Roth (most valuable to preserve). Before age 59.5, traditional withdrawals cost an extra 10% penalty, so Roth moves ahead of traditional in the order. Because withdrawing money to pay tax is itself taxable, the split is solved to a fixed point rather than estimated once.
For taxable sales, only the gain portion counts as income. Linked accounts use their reported cost basis; without one, we assume half of every dollar sold is gains, which deliberately overstates tax rather than understating it.
Required minimum distributions follow SECURE 2.0: age 73 if you were born 1951-1959, age 75 if born 1960 or later. Inherited (non-spouse) traditional money follows the 10-year rule and is spent before your own traditional dollars, since it is forced out soon anyway.
Your FIRE target is the larger of two numbers, and both come from the same month-by-month simulation of your actual plan. The withdrawal-rate target is the smallest balance that survives your plan if the portfolio earns only the real return your safe withdrawal rate implies over your horizon. The simulation target is the same walk at your plan's expected returns. Taking the maximum keeps the goal consistent with the projection chart: hitting your number always means the projected plan works.
Because it is your real plan rather than a spending average, the target prices what actually happens: daycare that ends, a mortgage that pays off, an inherited IRA that must empty on a deadline, Social Security that starts late, and the tax on each of them in the year it lands. Temporary costs are priced for the years they run instead of being treated as permanent, which is what a flat 25-times-expenses rule cannot do.
Reaching pre-tax money before 59½ is priced too. The plan compares modeled targets for a Roth conversion ladder, fixed 72(t) payments, and paying the early-withdrawal penalty. Rule of 55 requires an explicit selection and eligible employer separation. You can pin a route on the Strategy page. The goal card shows which route was chosen, why, and what the alternative would cost.
FIRE Journey is a planning tool, not financial, tax, or investment advice. The model makes honest simplifications. State and local tax is one effective rate per state rather than brackets. Itemized deductions are a single number you enter, so mortgage interest does not shrink as the loan amortizes. Every account earns the same return, since your plan has one allocation glide path rather than one per account. Health-insurance benchmarks come from a national age curve unless you override them. When no accounts are linked, your portfolio is assumed to be roughly 30% taxable, 50% traditional, and 20% Roth. Simulations draw from US market history even for non-US households, which use a single flat tax rate instead of the US engine.
Unknown account details require assumptions that can raise or lower your target. Unknown Roth contribution basis is treated as zero; verify account basis and withdrawal eligibility before relying on an early-retirement route. Subsidy cliffs use the costlier branch when the model cannot find a consistent premium. Future returns, inflation, and tax law will not match any model, and historical success rates describe the past rather than a guarantee. Treat every output as an estimate to stress-test, and talk to a professional before acting on it.