Historical cycles + Monte Carlo, side by side
How long does your money actually last?
Not a 4% rule of thumb — a real simulation run against 97 years of actual market history, or a randomized Monte Carlo stress test. Change any number below and the chart updates immediately.
Runs your scenario starting from every year in 1929–2025 (97 real market cycles), including the Depression, 2008, and 2022.
Withdraw the same inflation-adjusted amount every year, regardless of performance. This is the classic "4% rule" approach.
Success rate
Advertisement
How this works
Data sources & vintage
- Equity returns
- S&P 500 annual total returns, 1929–2025
- Bond returns
- 10-year US Treasury annual returns, 1929–2025
- Inflation
- US Consumer Price Index (CPI-U), December-to-December
- Coverage
- 97 complete years · all figures inflation-adjusted to real terms
- Last updated
- August 2026
Full source list and academic references are at the bottom of this page. The simulation runs entirely in your browser — the dataset is embedded in the page, so you can inspect it in view-source if you want to verify the numbers yourself.
Every simulation draws from the same base dataset: annual real (inflation-adjusted) total returns for the S&P 500 and 10-year Treasury bonds, 1929–2025, built from Robert Shiller's dataset and NYU Stern's historical returns series. Returns are real, so every dollar in this tool — your savings, spending, and results — is already expressed in today's purchasing power.
Historical Cycles
This mode runs your exact scenario starting from every one of the 97 years in the dataset, in the order they actually happened. If your retirement is longer than the years remaining in the data, the sequence wraps back to the beginning. This is the same method FIRECalc uses, and it's the standard for a reason: real market history has autocorrelation and fat-tail crashes that synthetic models tend to understate.
Monte Carlo
This mode runs 1,000 randomized trials, where each year's return is drawn independently, with replacement, from the same 97-year dataset (a technique called bootstrap resampling). It preserves the real shape of historical returns rather than assuming a bell curve, but breaks the actual chronological sequence — so it can generate harder back-to-back-bad-year sequences than ever actually occurred, which is why Monte Carlo results usually run a few points more conservative than Historical Cycles for the same inputs.
Withdrawal strategies
- Fixed real dollar — withdraw the same inflation-adjusted amount every year. The classic "4% rule" is this strategy at a 4% initial rate.
- % of portfolio — withdraw a fixed percentage of the current balance each year. Mathematically can't hit zero, but income varies with the market.
- Guardrails — a simplified version of the Guyton-Klinger decision rules. Withdrawals stay fixed in real terms unless the withdrawal rate drifts more than 20% above or below your starting rate, at which point spending is adjusted 10% to bring it back in line.
What's simplified
Social Security is entered as a flat monthly amount in today's dollars rather than calculated from earnings history — use the estimate from your Social Security statement. This tool doesn't model taxes, required minimum distributions, healthcare costs, or state-specific rules. It's a planning aid, not financial advice.
Reading the chart
The shaded bands show the spread of outcomes across every trial, not a single projection. The gold line is the median — half of all simulated retirements ended above it, half below. The inner band covers the 25th to 75th percentile, and the outer band stretches to the 10th and 90th. The dashed red line is zero: any path that touches it has run out of money.
A wide fan isn't a flaw in the model, it's the actual finding. Sequence-of-returns risk means two retirees with identical portfolios and identical average returns can end up in completely different places depending purely on when the bad years arrive. A crash in year two of a 30-year retirement is far more damaging than the same crash in year 25, because the early loss compounds against a portfolio you're simultaneously drawing down.
Why the two modes disagree
Historical Cycles usually reports a higher success rate than Monte Carlo on identical inputs, and that gap is informative rather than a bug. Real market history contains only 97 starting points, and the bad years within it arrived in a particular order — the 1929 crash, the 1970s stagflation, 2008, and 2022 each unfolded with their own recovery pattern.
Bootstrap resampling breaks that ordering. It can deal you three 1931-magnitude years back to back, which never actually happened. That produces a wider, harsher distribution of outcomes. Neither number is the "true" answer: historical cycles tell you what has happened, Monte Carlo tells you what the same building blocks could produce in a worse arrangement.
Common questions
What success rate should I aim for?
There's no universal threshold. Much published research targets around a 90% probability of not running out over a 30-year horizon, but a rate below 100% doesn't mean failure is likely — it means some historical or simulated paths ran short. What matters more is how much flexibility you have: a retiree who can cut spending in a bad year is in a very different position from one whose budget is entirely fixed.
Is the 4% rule still accurate?
It's a starting point rather than a settled number, and current research genuinely disagrees on the figure. William Bengen, who originated the rule in 1994, has revised his own worst-case estimate upward to roughly 4.7% using a more diversified portfolio, while Morningstar's forward-looking 2026 research puts a conservative fixed-spending baseline nearer 3.9%. Our safe withdrawal rate guide explains why those numbers differ.
Are the results in today's dollars?
Yes. Every figure — your savings, your spending, and every result — is in real, inflation-adjusted terms. The underlying return data has already been deflated by CPI, so you don't need to guess at future inflation or mentally discount the ending balances.
Why does my retirement length change the answer so much?
Because a longer horizon means more chances to encounter a bad sequence, and less time to recover from one. A 20-year retirement tolerates a meaningfully higher withdrawal rate than a 30-year one, and early-retirement horizons of 40 to 50 years are more demanding again — which is why FIRE planning often uses rates well below 4%.
How should I model Social Security if I haven't claimed yet?
Enter your estimated monthly benefit in today's dollars and set the year of retirement it begins. Because the simulation runs in real terms, you don't need to inflate the figure — the tool treats it as purchasing power that offsets your portfolio withdrawals from that year onward.
Does this account for taxes?
No. Taxes depend heavily on account type, state of residence, and withdrawal ordering, none of which this tool models. A common workaround is to enter your spending figure gross — that is, including the amount you expect to pay in tax — so the withdrawals being simulated reflect what actually leaves the portfolio.
Sources & references
Market data
- Robert J. Shiller, Yale University — long-run US stock market, dividend, earnings and CPI series. Used here for the inflation (CPI-U) series underlying every real-return calculation. Online data
- Aswath Damodaran, NYU Stern School of Business — annual returns on US stocks, Treasury bonds and Treasury bills, 1928–present. Used for the S&P 500 and 10-year Treasury return series. Current data
- US Bureau of Labor Statistics — CPI-U figures for years beyond the Shiller series. Consumer Price Index
Withdrawal rate research
- Bengen, W. P. (1994). "Determining Withdrawal Rates Using Historical Data." Journal of Financial Planning, October 1994. The paper that introduced the 4% rule and the SAFEMAX concept.
- Bengen, W. P. (2025). A Richer Retirement. Revises the author's own SAFEMAX estimate upward using a more broadly diversified portfolio.
- Guyton, J. & Klinger, W. (2006). "Decision Rules and Maximum Initial Withdrawal Rates." Journal of Financial Planning. The source of the guardrails strategy implemented in this tool.
- Cooley, Hubbard & Walz (1998). "Retirement Savings: Choosing a Withdrawal Rate That Is Sustainable." AAII Journal — commonly known as the Trinity Study.
- Morningstar (December 2025). The State of Retirement Income — annual forward-looking safe withdrawal rate research.
Comparable tools
The historical-cycle method used here follows the approach established by FIRECalc and cFIREsim, both long-running free retirement simulators. This tool differs mainly in adding bootstrap Monte Carlo alongside historical cycles, and in the withdrawal strategies offered.
About this tool
When to Retire is an independently built, ad-supported calculator. It is not affiliated with any brokerage, advisory firm, fund company, or financial institution, and it does not sell products, collect leads, or receive compensation for referrals. There is no account system and no data collection — every calculation runs in your browser and nothing you enter is transmitted or stored.
The simulation code, the underlying dataset, and the assumptions behind both are documented in full on this page rather than hidden behind a black-box score. If you find an error in the methodology or the data, that's worth knowing about — corrections can be sent to contact@whentoretire.net.
Important: This is an educational planning tool, not financial advice. It does not account for taxes, required minimum distributions, healthcare or long-term care costs, fees, annuities, or your individual circumstances, and past market performance does not predict future results. Retirement decisions carry real consequences — consider consulting a qualified financial professional, such as a fee-only fiduciary advisor, before acting on any projection produced here.