Century TestedRetirement income planner Check my plan

Methodology and data sources

How every Century Tested calculator works: the data behind it, the rules each strategy follows, and what the results can and can’t tell you.

By · Updated

Data sources

Every Century Tested result is built from four yearly U.S. series covering 1928 through 2025:

SeriesWhat it representsSource
StocksS&P 500 total return, dividends reinvestedAswath Damodaran, NYU Stern, “Historical Returns on Stocks, Bonds and Bills”
Bonds10-year U.S. Treasury bond total returnDamodaran, NYU Stern
Cash3-month U.S. Treasury bill return (average rate during the year)Damodaran, NYU Stern
InflationConsumer Price Index (CPI-U), December to DecemberU.S. Bureau of Labor Statistics

Figures are rounded to two decimals. The returns follow Damodaran’s January 2026 update, which includes final figures for 2025. Every calculation converts returns to real (after-inflation) terms, so all dollar amounts are in today’s money.

Historical testing: the Retirement Replay engine

Retirement Replay is the engine behind every Century Tested result. For a retirement of a given length, it tests every start year with complete data. A 30-year retirement has 69 possible start years, 1928 through 1996. Each test applies that stretch’s actual returns and inflation in order, year by year:

  1. At the start of each year, that year’s withdrawal is taken out (spending minus any Social Security or pension income you entered).
  2. What remains earns that year’s real return for your stock and bond mix, rebalanced yearly.
  3. A retirement “lasts” if every year’s withdrawal could be paid in full.

“Historical success” is the share of start years that lasted. Because consecutive start years share most of their market history, these results are not independent trials; they show how a plan would have held up through real events, not a probability of future success. For what these percentages mean in practice, see what success rate to aim for.

Simulated markets

Some comparisons also report results from simulated markets. Each simulated retirement is stitched together from random 5-year blocks of real history, keeping each year’s stock, bond and inflation figures together. Keeping years in blocks preserves real crash-and-recovery patterns that shuffling single years would break. The planner and the Monte Carlo vs historical page run 3,000 simulations and the strategy comparison pages run 2,000, each with a fixed random seed so results are repeatable. See Monte Carlo and historical results side by side on the same plan on that page.

Withdrawal strategies

Fixed spending (4% rule style)
The first year’s withdrawal rises with inflation every year and never changes otherwise. This is the approach tested in William Bengen’s 1994 study that gave rise to the 4% rule.
Guardrails
Based on the guardrail rules published by Jonathan Guyton and William Klinger (2006). Each year we compare your current withdrawal rate with your starting rate. If it has risen more than 20% above the start, spending is cut 10%; if it has fallen more than 20% below, spending is raised 10%. Cuts never take spending below a floor, 80% of starting spending by default.
Two buckets
A safety bucket holds the next 4 years of withdrawals (adjustable), invested half in Treasury bills and half in 10-year Treasuries. The rest is a growth bucket, 80% stocks by default. Withdrawals come from the safety bucket. It is refilled from growth only when the growth bucket is within 5% of its previous high, so refills don’t sell investments after a decline.
Guardrail buckets
Guardrails set each year’s spending, and the bucket pays it. One extra rule covers lean years: if growth investments fell last year and the cash bucket can’t cover the full plan, you spend a little less that year instead of selling. The cut is at most one guardrail step (10%) and never goes below your floor. Anything cash still can’t cover is sold from growth.

Investment mix

Unless you change it, fixed and guardrail plans hold 60% S&P 500 stocks and 40% 10-year Treasuries. Bucket plans keep four years of withdrawals in Treasuries and invest the rest 80% in stocks, which works out to a similar overall stock share for typical withdrawal rates. Portfolios are rebalanced once a year. Results are not based on a 100% stock portfolio.

How our numbers compare with published research

Two studies shaped how most people think about withdrawal rates. Asking our engine the same questions gives similar answers, running a little lower at 4%. Each comparison below lists how the methods differ.

Bengen (1994): where the 4% rule came from

William Bengen tested a portfolio of 50% stocks and 50% intermediate-term Treasuries, rebalanced, with withdrawals raised for inflation each year, using Ibbotson data from 1926. He found that a 4% first-year withdrawal never ran out in fewer than 33 years, while 4.25% could run out in as little as 28.

With the same 50/50 mix on our data, the highest rate that lasted 30 years in every start year since 1928 is 3.67% (3.73% with our default 60% stocks). At 4%, the 1966 retiree ran out in year 25.

How the methods differ: we use 10-year Treasuries rather than intermediate-term notes, and Damodaran and BLS data rather than Ibbotson’s. Bengen’s data ended in the early 1990s, so retirements that began in the late 1960s could not yet be followed for a full 30 years. Ours run through 2025, so they can.

The Trinity study: success rates

The study known as the Trinity study, by Philip Cooley, Carl Hubbard and Daniel Walz, tested withdrawals raised for inflation using S&P 500 stocks and long-term, high-grade corporate bonds from 1926 to 1997. The table shows the 30-year success rates from their 1999 paper, “Sustainable Withdrawal Rates From Your Retirement Portfolio” (Financial Counseling and Planning, vol. 10, Table 2), next to ours for the same rates and mixes.

Stocks / bondsRateTrinity (1926–1997)Century Tested (1928–2025)
100% / 0%3%100%69 of 69 (100%)
100% / 0%4%98%64 of 69 (93%)
100% / 0%5%81%54 of 69 (78%)
75% / 25%3%100%69 of 69 (100%)
75% / 25%4%100%65 of 69 (94%)
75% / 25%5%86%53 of 69 (77%)
50% / 50%3%100%69 of 69 (100%)
50% / 50%4%95%64 of 69 (93%)
50% / 50%5%70%46 of 69 (67%)

How the methods differ: Trinity had 43 overlapping 30-year periods, starting 1926 through 1968. We have 69, starting 1928 through 1996, which adds the retirements that began in the high-inflation years around 1970. Trinity withdrew monthly and held corporate bonds. We withdraw once a year, at the start of the year, and hold 10-year Treasuries.

All three tests agree on the broad picture: with a balanced mix, about 4% lasted 30 years in most or all of U.S. history, and 5% ran out in a meaningful share of start years. See every rate side by side in withdrawal rates tested.

Limitations

  • Taxes and fees are not included. Withdrawals from pre-tax accounts are taxed as income, and fund fees reduce returns.
  • U.S. history only. The U.S. had one of the strongest markets of the 20th century. Other countries’ retirees saw worse outcomes, and future U.S. returns may be lower.
  • Index returns. Real portfolios differ from the S&P 500 and 10-year Treasuries.
  • Yearly steps. We model one withdrawal and one return per year, not monthly cash flows.
  • Limited history. There are fewer than four independent 30-year periods since 1928. Results are a stress test, not a forecast.
  • Personal factors. Health, longevity, home equity, required minimum distributions and Social Security claiming rules all matter and are simplified here.

Disclosures

Century Tested is an educational tool. It is not a registered investment adviser, broker-dealer or tax adviser, and nothing on this site is a recommendation to buy or sell any security or to adopt any strategy. Results are hypothetical, based on historical index data, and do not represent actual investment results. Past performance does not guarantee future results. Before making decisions, consider consulting a qualified, fiduciary financial professional.