Hebeler Autopilot II Calculator

Hebeler’s Autopilot II: 75% of last year’s withdrawal raised for inflation, plus 25% of the payment amortising the balance over life expectancy.

Part of the FIRE simulator, a historical backtest calculator that replays your plan across every rolling window of market history. Backtests use US market data from 1871 through 2025.

How Hebeler Autopilot II did in US market history

Hebeler Autopilot II ran out of money in 0 of 126 start years of US market history from 1871 through 2025, on the simulator’s default plan: $1,000,000 in 80% stocks, 15% bonds and 5% cash, over 30 years, before fees and taxes, rebalanced yearly. Its first withdrawal was $40,000. Over the same 126 start years, Constant Amount, which takes $40,000 (4% of the portfolio) in the first year and raises it with inflation, ran out in 2 (the retirements that began in 1966 and 1969). Its leanest year came in 1985, in the retirement that began in 1956: $21,500 after inflation, 54% of that retirement’s first year. The worst start year was 1966, the retirement that paid out the least in total after inflation.

How the Hebeler Autopilot II strategy works

Hebeler Autopilot II withdraws the starting amount in year one. Each later year it takes 75% of last year’s withdrawal, raised by last year’s inflation, plus 25% of a planning amount: the payment that would spend the current balance to zero over the remaining life expectancy at that year’s age, assuming a 1.45% return after inflation. By default the plan starts at age 60.

This follows the rule Henry Hebeler published in 2015, with the one change named below. The planning amount alone follows the balance, so it would move spending sharply from one year to the next; anchoring three-quarters of each withdrawal on last year’s keeps income smooth while the planning quarter pulls it toward what the portfolio can support. In the US period life table for 2020-2024, remaining life expectancy at 60 is 23.1 years, and it shortens by less than a year each year. A period table freezes one span of years’ death rates, and real cohorts have kept outliving those, so it understates how long the money may need to last. The assumed return is Hebeler’s 5% less his 3.5% inflation. The change: life expectancy comes from that US table for both sexes, from the Human Mortality Database, where Hebeler used the IRS Publication 590 table. The period table gives a shorter span at retirement ages, so this version spends somewhat faster than his.

Where the Hebeler Autopilot II strategy comes from

Henry "Bud" Hebeler described the rule in a MarketWatch column of 16 January 2015: three-quarters of last year’s budget increased by inflation, plus one-quarter of a planning amount computed with a 5% return, 3.5% inflation and the life expectancies in IRS Publication 590.

  1. Hebeler, H. K. (2015). There’s a Better Way to Plan Retirement Withdrawals. MarketWatch, 16 January 2015. The link opens an Internet Archive copy.
  2. Human Mortality Database (2026). Period life tables. Revised 15 Jun 2026, published under CC BY 4.0.

What the Hebeler Autopilot II strategy trades off

Income is smooth and drifts slowly toward what the portfolio can support, paced to remaining life expectancy rather than to the plan’s length, so the start age moves the result a long way. On the default 30-year plan, with a 4% first-year withdrawal, started at 60, every historical window lasted. Started at 65, the same plan runs past the 19.4 years of life expectancy at that age, and most windows run out of money in the years after it. Life expectancy is an average, and many people alive at a given age live well past it.

Pros and cons

Pros

  • Smooth, stable year-to-year income
  • Spending paced to remaining life expectancy, not a fixed end date
  • Automatic, rules-based adjustment

Cons

  • Adapts slowly to sudden market moves
  • Paced to an average lifespan, which many retirees outlive
  • A later start age can turn a plan that lasts into one that runs out

Parameters you can adjust

  • Prior-spend weight (share of last year): How much of each year’s withdrawal is anchored to last year’s withdrawal, raised by last year’s inflation, versus a fresh planning amount. Higher (closer to 1) makes income smoother.
  • Assumed return after inflation: The yearly return, after inflation, that the planning part of the blend assumes when spreading the balance over the remaining life expectancy. The default is Hebeler’s 5% return net of his 3.5% inflation: 1.05 ÷ 1.035 - 1, about 1.45%.
  • Age at the start of the plan: The retiree’s age in the plan’s first year. Each year the planning part spreads the balance over the remaining life expectancy at that year’s age, from the US period life table for both sexes. An older start means a shorter span and larger withdrawals.

Questions about the Hebeler Autopilot II strategy

Is this Hebeler’s own Autopilot method?

Yes, with one difference. It uses his blend, his inflation raise on last year’s withdrawal and his return assumptions, but reads life expectancy from the US period life table for both sexes rather than IRS Publication 590. The US table gives a shorter span at retirement ages, so this version spends somewhat faster.

How is Hebeler Autopilot II different from the 4% rule?

The 4% rule raises a fixed amount with inflation for the whole plan. This rule builds 75% of each year’s withdrawal from last year’s, raised with inflation the same way, and lets the other 25% follow what the portfolio can support over the remaining life expectancy, so income drifts toward the balance slowly.

Why does the start age change Hebeler Autopilot II’s results?

Because the planning quarter spreads the balance over remaining life expectancy, an older start spends faster. On the default 30-year plan every historical window lasted from 60, and most ran out from 65, in the years after the plan passed the life expectancy at that age.

Backtest Hebeler Autopilot II against market history

Opens the simulator with Hebeler Autopilot II already selected, so you only set your portfolio and horizon.

Run the Hebeler Autopilot II simulation