Amounts are in today’s money. Your inputs stay in this browser.

In today’s money.

First-year withdrawal as a share of the portfolio.

How long the withdrawals must last.

US history resamples real years from 1928 to 2025; random returns draw each year independently.

Arithmetic mean of the yearly real return.

Standard deviation of the yearly real return.

More options

More simulations narrow the sampling noise; the model’s assumptions matter far more.

The same seed and inputs always give the same result. Change it to see how much is noise.

Taken out of each withdrawal before it is spent. Not a country tax estimate.

Middle of the year is the planner’s default. The worked example uses the formula’s convention.

For display only. Nothing is converted.

Over 1,000 simulated 30-year retirements from $1,000,000, constant real withdrawals of $40,000 ran short before the final year in 12.7% of them; a fixed 4% of the portfolio never ran out, but income fell below $30,000 at some point in 53%.

Four rules on the same simulated markets (income is what the rule paid, in today’s money)
RuleYear-1 incomeMedian income, year 1510th percentile, year 30Income ever below 75% of year 1Median largest one-year cutRan short before the final yearMedian left at the end
Constant real$40,000$40,000$013.7%0.0%12.7%$1,100,299
Fixed percentage$40,000$41,900$18,72653.1%20.6%0.0%$1,042,656
Guyton-Klinger guardrails$40,000$40,000$25,98227.0%10.0%0.2%$1,006,763
VPW$65,051$58,442$17,70173.0%23.8%0.0%$1,637

Constant real: yearly income

Constant real: yearly income$0$37.5K$75K$112.5K$150K51015202530Year of retirement75% of year 1
  • 10th to 90th percentile (lighter band)
  • 25th to 75th percentile (darker band)
  • Median
Constant real: median income starts at $40,000 and is $40,000 in year 30; in the worst tenth of simulations it is $0 or less in year 30.
Show the numbers as a table
Constant real: income percentiles by year (today’s money)
Year10thMedian90th
1$40,000$40,000$40,000
2$40,000$40,000$40,000
3$40,000$40,000$40,000
4$40,000$40,000$40,000
5$40,000$40,000$40,000
6$40,000$40,000$40,000
7$40,000$40,000$40,000
8$40,000$40,000$40,000
9$40,000$40,000$40,000
10$40,000$40,000$40,000
11$40,000$40,000$40,000
12$40,000$40,000$40,000
13$40,000$40,000$40,000
14$40,000$40,000$40,000
15$40,000$40,000$40,000
16$40,000$40,000$40,000
17$40,000$40,000$40,000
18$40,000$40,000$40,000
19$40,000$40,000$40,000
20$40,000$40,000$40,000
21$40,000$40,000$40,000
22$40,000$40,000$40,000
23$40,000$40,000$40,000
24$40,000$40,000$40,000
25$40,000$40,000$40,000
26$40,000$40,000$40,000
27$39,885$40,000$40,000
28$2,403$40,000$40,000
29$0$40,000$40,000
30$0$40,000$40,000

Fixed percentage: yearly income

Fixed percentage: yearly income$0$37.5K$75K$112.5K$150K51015202530Year of retirement75% of year 1
  • 10th to 90th percentile (lighter band)
  • 25th to 75th percentile (darker band)
  • Median
Fixed percentage: median income starts at $40,000 and is $41,421 in year 30; in the worst tenth of simulations it is $18,726 or less in year 30.
Show the numbers as a table
Fixed percentage: income percentiles by year (today’s money)
Year10thMedian90th
1$40,000$40,000$40,000
2$34,401$40,098$46,301
3$32,181$39,962$49,130
4$30,780$40,179$52,788
5$29,713$40,412$54,387
6$28,445$40,391$58,609
7$28,422$41,138$60,694
8$27,148$41,239$62,336
9$26,586$40,952$63,761
10$25,679$40,833$65,447
11$24,923$41,104$67,848
12$24,066$41,045$69,530
13$24,006$41,398$70,801
14$23,261$41,655$71,845
15$23,252$41,900$73,583
16$22,318$41,745$76,798
17$22,023$41,513$78,064
18$21,403$41,490$78,697
19$20,541$42,056$79,986
20$20,798$42,320$80,863
21$19,995$42,235$83,978
22$20,194$42,514$84,871
23$20,218$41,967$86,194
24$19,695$42,243$88,928
25$19,496$42,394$91,032
26$18,835$42,157$94,036
27$19,079$41,421$97,284
28$18,906$42,019$97,686
29$18,893$42,136$100,635
30$18,726$41,421$102,455

Guyton-Klinger guardrails: yearly income

Guyton-Klinger guardrails: yearly income$0$37.5K$75K$112.5K$150K51015202530Year of retirement75% of year 1
  • 10th to 90th percentile (lighter band)
  • 25th to 75th percentile (darker band)
  • Median
Guyton-Klinger guardrails: median income starts at $40,000 and is $48,400 in year 30; in the worst tenth of simulations it is $25,982 or less in year 30.
Show the numbers as a table
Guyton-Klinger guardrails: income percentiles by year (today’s money)
Year10thMedian90th
1$40,000$40,000$40,000
2$40,000$40,000$40,000
3$36,000$40,000$40,000
4$36,000$40,000$44,000
5$32,400$40,000$44,000
6$32,400$40,000$48,400
7$32,400$40,000$53,240
8$29,160$40,000$53,240
9$29,160$40,000$53,240
10$29,160$40,000$58,564
11$26,244$40,000$58,564
12$26,244$40,000$63,841
13$26,218$40,000$64,420
14$23,620$40,000$64,420
15$23,620$40,000$70,862
16$23,620$40,000$70,862
17$23,620$40,000$70,862
18$23,620$43,124$70,862
19$23,620$43,560$77,949
20$23,620$43,560$77,949
21$23,620$43,560$77,949
22$23,620$44,000$85,744
23$23,620$44,000$85,744
24$25,696$45,014$85,744
25$25,696$47,676$93,375
26$25,722$47,916$94,318
27$25,956$47,916$94,318
28$25,982$48,400$94,318
29$25,982$48,400$102,712
30$25,982$48,400$103,750

VPW: yearly income

VPW: yearly income$0$37.5K$75K$112.5K$150K51015202530Year of retirement75% of year 1
  • 10th to 90th percentile (lighter band)
  • 25th to 75th percentile (darker band)
  • Median
VPW: median income starts at $65,051 and is $42,158 in year 30; in the worst tenth of simulations it is $17,701 or less in year 30.
Show the numbers as a table
VPW: income percentiles by year (today’s money)
Year10thMedian90th
1$65,051$65,051$65,051
2$55,233$64,518$74,635
3$51,057$63,605$78,415
4$48,288$63,261$83,426
5$46,068$62,953$85,073
6$43,609$62,237$90,805
7$43,087$62,728$93,065
8$40,668$62,187$94,523
9$39,352$61,101$95,607
10$37,584$60,202$97,228
11$36,058$59,909$99,706
12$34,404$59,198$101,213
13$33,805$59,075$102,022
14$32,411$58,709$102,262
15$32,069$58,442$103,282
16$30,330$57,458$107,015
17$29,597$56,477$107,385
18$28,343$55,733$107,260
19$26,864$55,898$108,276
20$26,794$55,744$107,601
21$25,444$54,506$110,770
22$25,378$54,184$109,930
23$25,052$52,992$111,110
24$23,887$52,572$113,334
25$23,383$51,956$114,572
26$22,171$51,143$116,877
27$22,045$49,205$119,136
28$21,004$49,169$118,899
29$20,368$47,773$120,046
30$17,701$42,158$110,246

How this was calculated

  • Random returns: each year’s real return is drawn independently from a log-normal distribution with an average of 5% and a standard deviation of 12% (the engine’s “normal” sampler with a log-normal shape). Inflation is held at 0, so every figure is in today’s money.
  • 1,000 simulations per rule with seed 2026; every rule sees the same market sequences.
  • Rules (the engine’s withdrawal strategies): constant real takes 4% of the starting portfolio, then raises it with inflation; fixed percentage takes 4% of each year’s balance; Guyton-Klinger starts at 4%, skips the inflation raise after a losing year when the rate is above the initial one, cuts 10% when the rate is over 120% of the initial rate (not in the last 15 years) and raises 10% when it is under 80%; VPW amortises the balance over the years left at an assumed real return (5% for stocks, 1.9% for bonds, weighted by the mix; the random-returns model counts as all stocks).
  • “Ran short before the final year” counts simulations with a year before the last in which the rule’s amount could not be paid in full (14.1% of constant-real simulations had any short year, including the last).
  • Money moves in the middle of each year (half a year’s growth), the planner’s default.
  • No tax. The engine’s flat tax mode is set to 0%; the full planner estimates tax for 12 countries.

Results update as you type (simulations run in the background and can be cancelled). The first result shown is for the default inputs, calculated when this page was built.

The four rules

  • Constant real (“the 4% rule”). Year 1 takes your initial rate of the starting portfolio; every later year takes the same amount raised by inflation. The portfolio’s value is never consulted. This is the rule inside William Bengen’s 1994 study.
  • Fixed percentage. Every year takes the same percentage of the current portfolio. It cannot run out, and income moves with the portfolio: a 25% fall in value is a 25% fall in next year’s income.
  • Guyton-Klinger guardrails, from Guyton’s 2004 and Guyton and Klinger’s 2006 papers. Withdrawals rise with inflation, except after a year with a negative return when the current rate is above the initial rate. If the current rate is more than 20% above the initial rate, spending is cut by 10% (not in the last 15 years); if it is more than 20% below, spending is raised by 10%.
  • VPW (variable percentage withdrawal). Each year takes the payment that would spend the portfolio exactly over the years left at an assumed return, so the percentage rises as the horizon shortens. The engine assumes 5% a year after inflation for stocks and 1.9% for bonds, weighted by the mix (the random-returns model counts as all stocks). The community version is described on the finiki VPW page.

How the calculator works

Each rule is set as the plan’s withdrawal strategy in the engine and run through the same Monte Carlo simulations: the same seed, so every rule sees exactly the same sequences of returns. “Income” is what the rule actually paid each year, in today’s money. The calculator then reads, for each rule, the first-year income, the middle and bad-luck income later on, how often income fell below 75% of its first year, the typical largest one-year cut, and how often the rule could not pay its own amount.

Two market models are offered: random real returns drawn independently each year from a log-normal distribution (average 5%, volatility 12% by default, the model in the blog article), or US history from 1928 to 2025 with random start years and a stock and bond mix.

Worked example

Computed by the calculator’s own code when this page was built: $1,000,000, a 4% initial rate, 30 years, random real returns averaging 5% with 12% volatility, 2,000 simulations with seed 2026, withdrawals at the start of each year, no tax.

Four rules on the same 2,000 simulated retirements (today’s money)
RuleYear-1 incomeMedian, year 1510th percentile, year 30Income ever below 75% of year 1Median largest one-year cutA year not paid in full
Constant real$40,000$40,000$012.8%0.0%13.4%
Fixed percentage$40,000$41,634$18,76253.4%20.2%0.0%
Guyton-Klinger guardrails$40,000$40,000$26,24426.2%10.0%0.5%
VPW$65,051$57,395$17,46775.6%22.3%35.7%

The pattern is the one the withdrawal strategies article found with its own script: constant real income never moves until it fails (in 13.4% of these simulations a year went unpaid); a fixed percentage never runs out but often cuts income; guardrails cap the typical cut at 10%; VPW starts higher and spends the portfolio down. For VPW, “a year not paid in full” is mostly its final year: the engine asks for the whole remaining balance grown at the assumed return, and a final year below that return leaves a small part unpaid.

What this assumes and leaves out

  • The results depend on the market model: a different return, volatility or horizon changes every row. Independent yearly returns understate runs of bad years; the history model keeps them but has few independent sequences.
  • One portfolio, rebalanced yearly; no fees; no tax unless you set the flat rate.
  • No pension or other income. A floor of guaranteed income changes which rule suits you most.
  • The Guyton-Klinger rules are applied as the engine implements them (one asset mix, so the portfolio management rule does not apply).
  • History data: Aswath Damodaran, Historical Returns on Stocks, Bonds and Bills (NYU Stern), and US Bureau of Labor Statistics CPI-U.

Questions

What is the best retirement withdrawal strategy?

There is no best rule, only a trade-off. Constant real withdrawals keep income steady and risk running out; a fixed percentage never runs out and lets income move with the market; guardrails and VPW sit between. Which suits you depends on how much of your spending is fixed, what other income you have and whether you want to leave money behind.

Is Guyton-Klinger the same as guardrails?

Guyton and Klinger described two of their rules as guardrails, and the name is now used for any rule that changes spending when a trigger is crossed. The full planner also has risk-based guardrails that react to the plan’s chance of success instead of its withdrawal rate.

Can a percentage-of-portfolio rule fail?

It cannot reach zero, because each withdrawal is a share of what is left. But income can fall to a level that does not cover your spending, which is a failure of a different kind; the column for income below 75% of year 1 measures that.

Why do the numbers differ from the blog article?

The article’s table came from an independent script with its own random draws, start-of-year withdrawals and a simpler VPW. This calculator runs the planner’s engine. The pattern is the same; individual figures differ by a point or two, and VPW starts higher because the engine’s VPW assumes a higher return.