A result you can trace back to its assumptions.

1. Keep inflation consistent

In the FIRE, Coast FIRE and compound-interest calculators, enter annual return before inflation. We convert it to a real return using:

real return = (1 + nominal return) ÷ (1 + inflation) − 1

At 7% return and 3% inflation, the result is 3.8835%, not exactly 4%. Spending, balances and contributions in these models are expressed in today’s purchasing power. A constant real contribution means the cash amount you invest would rise with inflation. If your assumption is already a real return, set inflation to zero.

2. How savings grow

We convert an effective annual return to a monthly rate: (1 + annual return)^1/12 − 1. Each month, the opening balance earns that return, then the contribution is added. The focused savings tools do not model withdrawals during saving. The complete planner also subtracts additional expenses or one-time events during this period.

next balance = current balance × (1 + monthly rate) + contribution

The main FIRE planner increases contributions after each completed year by the additional growth percentage you enter. This is growth above inflation. The simple spending-target result finds the first monthly balance reaching that target, with an 80-year limit. The complete timeline continues through the plan end age and includes the income and expenses you enter. Reaching the target is not proof the withdrawal plan will survive.

3. A withdrawal target is not a return forecast

target portfolio = annual spending ÷ withdrawal rate

At 4%, annual spending of 40,000 implies 1,000,000 invested. The classic fixed-real withdrawal approach starts with a percentage of the initial portfolio, then adjusts the cash withdrawal for inflation. Taking 4% of the current balance every year is a different strategy with variable income.

William Bengen’s 1994 paper examined historical US portfolios. Our target calculator only divides spending by a chosen rate; it does not reproduce his portfolio study. Read the original paper and our worked comparison of withdrawal rates.

4. Coast FIRE has two different dates

Your retirement age is the date withdrawals begin. Your coast age is the earlier date when contributions could stop, while work or another income still pays your living costs. We discount the retirement target back to each month using the real return. The first month your projected savings exceed that month’s coast threshold is the result.

Further contributions are included only until that threshold is reached. The separate “portfolio at retirement if saving continues” result assumes contributions continue all the way to retirement, for comparison.

5. The separate stock-only historical tool

The historical test uses 1928–2025 US large-cap stock returns including dividends, with US inflation from Aswath Damodaran’s NYU dataset. Earlier years use the source author’s predecessor-index series. This model is 100% stocks; it does not contain bonds or reproduce a balanced portfolio.

A full year’s spending is withdrawn at the beginning of each year. The remaining balance earns that year’s real return. All requested withdrawals must be funded; ending at exactly zero after funding the last withdrawal counts as survival. Overlapping historical windows share years and are not independent estimates of future probability.

Download the annual data (CSV) · Source details and calculation inputs. Retrieved 27 September 2026. Returns exclude your taxes and fees. Choosing a currency changes the monetary scale, not the underlying US market or inflation history.

6. The separate Monte Carlo tool

The simulator draws independent monthly normal returns. The monthly mean is (1 + annual mean)^1/12 − 1; volatility is divided by √12. Returns cannot take the portfolio below zero. Monthly spending is withdrawn first and rises at the fixed inflation assumption.

Identical inputs use the same random sample so you can compare changes without a new draw moving the answer. Results and chart balances are nominal future money. Chart percentile bands use a subset of runs; the reported ending values and survival percentage use all runs. The survival percentage describes these simulations, not the probability that your actual retirement will succeed.

The model does not reproduce market crashes, changing correlations, fees, taxes or uncertain inflation. Normal independent draws can understate real-world risk.

7. Currency, tax and saved plans

Currency conversions use the site’s configured exchange-rate snapshot or fallback rates. They are a convenience for comparing amounts, not a trading quote. A currency selection does not change tax laws or pension rules. Tax tools are simplified, country-specific estimates and are not included in the general FIRE projection.

Saved plans are stored in your browser. Clearing browser data can remove them. Optional analytics are governed by the choices on the privacy page.

8. The complete planner: model 2.0.0

The main planner keeps every amount in today's purchasing power. With an annual fee, real growth is (1 + nominal return) × (1 − fee) ÷ (1 + inflation) − 1. During saving, regular living costs are already outside the monthly contribution; added expense events are additional. Pension and extra income are added to invested cash during saving.

Retirement income first covers spending. Any surplus is invested. Any remaining expense is withdrawn at the beginning of the month; the remaining portfolio then earns that month's return. Pension and changed spending begin at the entered age. Recurring events end before their stated end age. One-time events occur once at the nearest modeled month. Ages and the plan horizon are resolved to months.

A full spending plan is funded only when every expense is met. The percentage strategy can keep the portfolio above zero by cutting regular spending; those cuts still count as shortfalls against the entered budget. Taxes are excluded. The optional retirement-age search uses constant returns and does not certify retirement safety.

Historical portfolio test

The main planner adds the same NYU source's nominal 10-year Treasury and three-month Treasury bill series. It resets the selected stock, bond and cash weights annually. Asset returns and historical inflation are converted to monthly equivalents within each year; observed month-by-month returns are not available in this dataset. The portfolio fee is deducted. Contributions before retirement follow the fixed-return assumption, and only the retirement period is stress tested.

Simulation test

The main planner draws 1,000 paths with a separate deterministic seed for each path. Monthly portfolio returns follow a normal distribution, with mean (1 + annual return)^1/12 − 1 and volatility divided by √12. Fees and the chosen inflation assumption are then applied. All main-planner outputs remain in today's money. The historical allocation does not automatically estimate the expected return or volatility; those are separate inputs. These conventions differ from the separate simulator's nominal-money display.

Independent normal draws omit changing correlations, market regimes and inflation shocks. Historical windows overlap and represent one market. Neither output is a personal probability of success. Inspect reproducible comparisons and complete data.

9. Tax models: nominal basis, real spending

Tax models use nominal account growth and nominal remaining cost basis. Month one's entered spending is withdrawn at month end; subsequent targets increase monthly with inflation. Year-end balances and tax payments can separately be discounted by the number of completed months. A taxable nominal gain is not removed merely because inflation reduces its purchasing power.

The ASK model withdraws remaining deposits, then available unused shielding, before taxable gains. Future shielding is credited at year end using the lowest remaining deposit balance plus unused shielding brought into that year; it becomes available for subsequent withdrawals. This conservative timing does not model same-year tax settlement. Future shielding defaults to 0% as a placeholder. The US tool assumes pro-rata basis recovery and a flat user-entered gain tax rate. Neither tool infers basis from market value, and neither models all tax rules. Sources and specific exclusions appear beside each calculator.

10. Verification and changes

Model 2.0.0, 27 September 2026: combined accumulation and retirement cash flow; pension, spending phases and life events; explicit fees; historical stocks, bonds and bills; saved scenarios, comparison and export; revised nominal tax-basis handling. The former dashboard now opens the same complete plan.

Automated checks cover closed-form compounding, zero and negative real returns, the first target month, exact final withdrawals, annual rebalancing, pension timing, irregular costs, nominal tax basis and preservation of the main plan. Published research is regenerated from the same engine and checked for reproducibility within numerical rounding tolerance. Browser checks cover entry, saving, sharing and mobile layout. These checks verify implementation behavior; they are not an independent audit or assurance about future markets.

FourPercent uses AI assistance for code and explanatory drafts. Financial rules are linked to primary sources; examples are computed from the stated models. No certified financial or tax review is claimed. If you find an error, use the correction contact and include the relevant assumptions.