Advanced FIRE Calculator

Monthly backtesting with 1,863 months of Shiller data (1871-2026), four withdrawal strategies compared side-by-side, Social Security and pension modeling, Monte Carlo simulation with stock-bond correlation, and fee drag analysis.

$
$
years
years
%
%

Social Security / Pension
$
years
$
years

Spending Changes
Model the well-documented retirement spending decline. Research by David Blanchett (Morningstar) shows real spending typically drops 1-2% per year after age 65, with a sharper decline after 80.
years
%

Constant Dollar
Success-- Median-- Worst-- Avg WD-- Cycles--
CAPE-Based
Success-- Median-- Worst-- Avg WD-- Cycles--
VPW
Success-- Median-- Worst-- Avg WD-- Cycles--
Guyton-Klinger
Success-- Median-- Worst-- Avg WD-- Cycles--
Success Rate
--
Worst End Balance
--
Median End Balance
--
Best End Balance
--
Worst Start Year
--
Fee Drag Impact
--
Survived Failed Median Path

Historical Cycles

StartEnd BalanceMin BalanceAvg WithdrawalResult
Success Rate
--
5th Percentile
--
25th Percentile
--
Median End
--
75th Percentile
--
95th Percentile
--
10,000 simulations using monthly arithmetic mean returns and standard deviations from the full Shiller dataset. Stock-bond correlation (r = 0.18) modeled via Cholesky decomposition. Fee drag deducted monthly.
All Paths Median 5th/95th
VPW table computed using Bogleheads methodology: terminal age 100, blended real discount rate of 4.23% for your 75/25 allocation. Verified against official Bogleheads VPW tables.
AgeYears LeftVPW %On $1M Portfolio

Why We Built This

Most FIRE calculators give you a single number and call it a day. Plug in your savings rate, pick 7% returns, and it spits out "17 years to FIRE." That's fine as a starting point, but it's not how retirement actually works. Markets don't return 7% every year. Sometimes they crash 35% right after you retire. Sometimes inflation eats your purchasing power for a decade. The 4% rule works until it doesn't, and the difference between "works" and "doesn't" often comes down to the specific 30-year window you happened to retire into.

This calculator exists because we wanted something that takes the question seriously. It runs your plan against every possible retirement start date in 155 years of U.S. market history, at monthly resolution. It lets you compare four different withdrawal strategies side by side so you can see the actual tradeoffs, not just read someone's opinion about them. And every number traces directly back to Robert Shiller's dataset at Yale, which you can download and check yourself.

Where the Data Comes From

Everything here is built on Shiller's ie_data.xls, the same dataset behind the CAPE ratio and most serious retirement research since the 1990s. Stock returns come from his Real Total Return Price index, which tracks the S&P Composite with dividends reinvested and CPI inflation stripped out. Bond returns come from his Real Bond Total Returns index for 10-year Treasuries. We computed month-over-month returns from these indices and cross-checked the results against Aswath Damodaran's independent dataset at NYU Stern. The two series correlate at r = 0.97 for stocks and r = 0.97 for bonds. The small differences come from Shiller using monthly average prices while Damodaran uses Dec 31-to-Dec 31 calendar year returns. Neither is wrong. They just measure slightly different things.

The Four Withdrawal Strategies

Constant Dollar is the classic. You withdraw a fixed real amount every year regardless of what the market does. It's what Bengen tested in 1994 and what the Trinity Study validated. Simple, predictable, and it works most of the time at 4%. The problem is that "most of the time" means a few unlucky cohorts (looking at you, 1966 retirees) get crushed by sequence-of-returns risk while everyone else dies with more money than they started with.

CAPE-Based adjusts your spending to how expensive the market is. When stocks are overpriced (CAPE above 30), you spend less. When they're cheap (CAPE below 15), you can spend more. The math is simple: withdrawal rate = 1/CAPE, clamped between a floor and ceiling you pick. It's elegant in theory and it does improve success rates, but it requires you to actually cut spending when the market is down, which is psychologically harder than it sounds.

VPW (Variable Percentage Withdrawal) is probably the most underappreciated strategy in the FIRE world. Developed by the Bogleheads community, it gives you a percentage that increases with age and varies by allocation. At 65 with a 60/40 portfolio, it says withdraw 5.0% of your current balance. At 80, that rises to 6.9%. It can never fully deplete your portfolio early, but it can mean sharp spending cuts in bad markets. The VPW percentages in this calculator match the official Bogleheads tables exactly. We verified all four allocation columns to the tenth of a percent.

Guyton-Klinger Guardrails starts like the constant dollar approach but adds rules for when things go off track. If your withdrawal rate drifts 20% above your initial rate (meaning the portfolio shrank), you cut spending by 10%. If it drifts 20% below (the portfolio grew), you give yourself a 10% raise. The capital preservation rule is removed 15 years before the end of your horizon, which is the original Guyton-Klinger design. It allows higher initial withdrawal rates than the 4% rule, often 5%+, but the spending cuts during bad markets can be steep.

Monthly vs. Annual Backtesting

Most free FIRE calculators run annual backtests. We run monthly, using all 1,863 data points in the Shiller dataset. Why does this matter? Because sequence-of-returns risk plays out month by month, not year by year. A retiree who withdraws at the start of January 2008 and watches the market fall 35% over the next 14 months has a very different experience than one who simply sees "-35.65%" as an annual return. Monthly resolution captures the compounding effect of withdrawing from a declining portfolio more accurately.

What the Monte Carlo Adds

Historical backtesting is limited to what actually happened. That's powerful, but it's also only about 125 independent 30-year windows. Monte Carlo generates 10,000 random scenarios from the same statistical distribution, which means it can show you outcomes worse than anything in recorded history. We model stock-bond correlation at r = 0.18 (the actual historical figure from 155 years of data) using Cholesky decomposition, so the simulated stock and bond returns have the right relationship to each other. The catch is that Monte Carlo assumes returns follow a normal distribution, which slightly understates the probability of extreme crashes.

Why Fee Drag Gets Its Own Section

Because it matters more than most people think. A 0.10% expense ratio, which is what you pay for a Vanguard Total Stock Market index fund, barely dents your outcome. But a 1.0% advisory fee, which is still shockingly common, can reduce your terminal wealth by 20-25% over 30 years. That's not a typo. This calculator shows you the exact impact by running your scenario with and without fees and comparing the median result. If that doesn't motivate you to check your expense ratios, nothing will.

Social Security and the Spending Smile

Real retirement spending doesn't stay flat. Research by David Blanchett at Morningstar found that retirees tend to spend less as they age, not more. The "spending smile" shows high spending in early retirement (travel, hobbies), a dip in the middle years, and a potential uptick for healthcare costs late in life. This calculator lets you model a spending reduction at a specific age to capture that pattern.

Social Security also changes the math significantly. If you're withdrawing $40,000/year from your portfolio and then Social Security kicks in at $2,000/month, your required portfolio withdrawal drops to $16,000/year. That's a huge reduction in sequence-of-returns risk. The calculator models SS and pension income as offsets to your withdrawal need, with the option to mark pensions as non-COLA (meaning their real value erodes over time).

Known Limitations

This calculator uses U.S.-only stock and bond data. If you hold international equities, TIPS, real estate, or alternatives, the historical backtest won't capture their behavior. The Monte Carlo assumes normally distributed returns, which understates tail risk. The current CAPE ratio (above 30 as of early 2025) is higher than almost all historical starting points, which means forward-looking returns may be lower than the 8.6% real average in the dataset. Past performance does not guarantee future results.

Why Taxes Are Not Included

This is deliberate. Tax drag on retirement withdrawals depends on whether your money is in a traditional 401(k) (taxed as ordinary income), a Roth (tax-free), or a taxable brokerage account (capital gains rates that depend on cost basis). It also depends on your state (nine states have no income tax), your filing status, your other income, and whatever tax law happens to be in effect 20 years from now. Modeling any of that generically would give you a precise-looking number that's almost certainly wrong for your situation. FICalc, cFIREsim, and FireCalc all exclude taxes for the same reason. If you want to account for taxes, the simplest approach is to gross up your spending number by your expected effective tax rate before entering it here.

Frequently Asked Questions

Where does the data come from?
Robert Shiller's ie_data.xls at Yale University. Stock returns use his Real Total Return Price index (column 9), bond returns use his Real Bond Total Returns index (column 18), and CAPE uses column 12. All returns are CPI-adjusted. You can download the same file and verify every number. We also cross-checked against Damodaran's dataset at NYU Stern: the two series correlate above 0.97 for both stocks and bonds.
How does Guyton-Klinger actually work?
You start with an initial withdrawal rate (say 5%). Each year you adjust for inflation, same as the constant dollar method. But if your current withdrawal rate drifts 20% above your initial rate (portfolio is shrinking), you cut your dollar withdrawal by 10%. If it drifts 20% below (portfolio is growing), you give yourself a 10% raise. The capital preservation rule stops applying 15 years before the end of your horizon, per the original 2006 paper. It nearly eliminates portfolio depletion, but the spending cuts in bad markets can be painful.
Which withdrawal strategy is best?
It depends on what you value. If you want predictable income, constant dollar is simplest. If you want to maximize total spending, VPW is hard to beat. If you want a higher initial withdrawal rate with guardrails, Guyton-Klinger allows that. If you want spending that adapts to market valuations, CAPE-based is elegant. Use the "Compare All Strategies" option to see the actual tradeoffs with your specific numbers. There is no universally best answer.
Why don't the results match cFIREsim or FireCalc exactly?
Different tools compute returns differently from the same Shiller data. Shiller's prices are monthly averages of daily closes, not month-end prices. Some tools use Jan-to-Jan returns, others use Dec-to-Dec. Some annualize differently. For marginal cases like the 1966 cohort at exactly 4%, a 1-2 percentage point difference in success rate is normal across tools. The statistical properties (means, standard deviations, correlations) are nearly identical across all reputable tools.
Is the 4% rule still safe?
With a 75/25 stock/bond portfolio and no fees, the 4% constant dollar rule has a 97.6% historical success rate over 30 years in this dataset. That's good but not perfect. The three failure cohorts all retired in the mid-to-late 1960s and got hit by a decade of stagflation. Whether a similar scenario could happen again depends on factors no backtest can predict. If you want more certainty, use 3.5%, or switch to VPW or guardrails, which adapt to market conditions.
How accurate is the Monte Carlo?
It generates 10,000 random return sequences using the actual mean, standard deviation, and correlation from 155 years of data. The stock-bond correlation (r = 0.18) is modeled properly via Cholesky decomposition. The main weakness is that real returns have slightly fatter tails than a normal distribution, so the Monte Carlo slightly understates the probability of extreme crashes. Use it alongside the historical backtest, not instead of it.
Data Sources and Verification: All stock and bond returns in this calculator are computed from Robert J. Shiller's ie_data.xls dataset at Yale University. You can download that file and verify every number yourself. Monthly and annual real stock total returns are computed from Shiller's Real Total Return Price index (column 9), which incorporates dividend reinvestment and CPI adjustment. Annual real bond total returns are computed from Shiller's Real Bond Total Returns index (column 18). CAPE values are the January reading for each year (column 12). Annual returns are January-to-January. Monthly returns are computed month-over-month from the same indices. Stock returns were cross-validated against Aswath Damodaran's Historical Returns (1928-2024) at NYU Stern: the two series correlate at r = 0.972 (stocks) and r = 0.967 (bonds), with divergences attributable to Shiller's use of monthly average prices versus Damodaran's Dec 31-to-Dec 31 calendar year returns. VPW percentages are computed using the methodology documented in the Bogleheads VPW wiki and verified against the official VPW tables for all four published allocation columns (20/80, 40/60, 60/40, 80/20).
Disclaimer: This calculator is for educational and informational purposes only and does not constitute financial, investment, or retirement planning advice. Past performance does not guarantee future results. The historical data reflects U.S. market conditions only. Consult a qualified financial advisor for personalized retirement planning. No representation is made as to the accuracy or completeness of any modeling or scenario analysis.