Retirement Withdrawal Simulator

What a Monte Carlo simulation does

A normal retirement calculator uses one average return, say 7% a year, every year. That produces a smooth line and a clean answer, but real markets do not behave that way. Some years are up 20% and some are down 15%, and the order in which those years arrive matters a great deal once you start taking money out. A Monte Carlo simulation replaces the single smooth line with thousands of different possible futures.

Each scenario here draws a random return for every year of your retirement, takes your spending out at the start of the year, lets the remainder grow or shrink by that year's return, and repeats until the end of the period or until the money runs out. The headline number is simply the share of scenarios in which the money lasted. It is a way of seeing risk, not a prediction: a "78% success rate" does not mean you have a 78% chance of anything in real life, it means that under the assumptions you chose, 78 out of every 100 random paths survived.

The tool re-runs with fresh random numbers each time you press the button, so the success rate moves by a percentage point or so from run to run. That wobble is itself informative: if two plans differ by less than that, the simulator cannot tell them apart.

How to read the chart and the numbers

• Success rate: the share of scenarios where the balance was still above zero after your last year. A scenario that hits zero in year 24 of 30 counts as a failure even though it lasted most of the way.
• Median balance: half of the scenarios ended above this number and half below. It is often much larger than your starting savings, which shows how lopsided the outcomes are: good markets leave a large surplus, bad markets leave nothing.
• 10th percentile: the balance that only the worst 10% of scenarios fall below. If this is zero, more than one scenario in ten ran out of money.
• Fan chart: the bands widen over time because uncertainty compounds. The thing to watch is how close the lower band comes to zero and when.

Example results

With the default assumptions (7% average return, 12% volatility, 2.5% inflation, 10,000 scenarios, a $1,000,000 starting balance and spending raised with inflation), the simulator gave these success rates. They are the output of this tool's own code, not historical data.

Two lessons stand out. First, a longer retirement costs a lot of success: the same 4% that survived 82% of 30-year scenarios survived only 54% of 50-year ones, which matters for anyone planning to retire in their 40s. Second, the relationship is steep. Moving from 4% to 5% drops the 30-year result from 82% to 58%.

The sensitivity to the settings is just as important. Holding the 4% and 30-year case fixed, changing the volatility from 8% to 20% moved the success rate from 95% down to 59%. Changing the average return from 5% to 8% moved it from 56% to 90%. Changing inflation from 2% to 4% moved it from 86% to 62%. The assumptions you type in matter more than the third decimal of any output.

Spending guardrails

Most failures in a simulation share a pattern: a bad market early on shrinks the balance while the plan keeps spending the same inflation-adjusted amount, so the withdrawal rate climbs and the portfolio cannot recover. Real retirees do not usually behave like that. They notice and trim spending.

The guardrails option models a simplified version of that behaviour. Each year it compares your current withdrawal rate (this year's spending divided by the balance) with the rate you started at. If it has risen by more than the upper threshold, spending is cut by the adjustment size, down to a floor. If it has fallen by more than the lower threshold, spending is raised by the same amount. This is a simplified rule of the guardrail type, not the full published Guyton-Klinger rule set, which has additional conditions.

In one run with a 5% starting rate over 30 years, the success rate was 58% without guardrails. Turning on guardrails with 20% thresholds, 10% steps and a 70% floor lifted it to about 85%. That is not free: in many of those scenarios spending ended at the floor, $35,000 instead of $50,000 in today's money. Guardrails convert the risk of running out into the risk of having to live on less, which for many people is the better trade, but you should look at the spending floor and ask whether you could really live on it.

Using the simulator well

• Test your own assumptions, not just the defaults. The defaults are moderately cautious for a balanced stock and bond portfolio. A more stock-heavy portfolio would justify a higher average return and a higher volatility together; changing one without the other is the common mistake.
• Subtract guaranteed income. If Social Security or a pension will cover $20,000 of a $60,000 budget, enter $40,000 as the spending to be funded from savings. See the Social Security break-even tool for the claiming decision.
• Think in terms of ranges. A plan at 90% is not "safe" and a plan at 75% is not "doomed". Ask what you would do in the bad scenarios, and whether you could do it.
• Pair it with the other tools. Use the FIRE calculator to find a target, then stress-test it here. Read the guides on sequence of returns risk and withdrawal strategies for the reasoning behind the results.

Frequently asked questions

Why does the success rate change when I run it again?

Each run draws new random scenarios. With 5,000 of them, the result typically moves by about a percentage point. Treat differences smaller than that as noise.

Is a higher success rate always better?

Not necessarily. A 99% success rate often means you are spending far less than you could, and the median ending balance will be very large. Many people aim for a result they are comfortable with and plan to adjust spending if markets go badly.

Why does the simulator disagree with the 4% rule?

The 4% rule comes from historical United States data, in which stocks returned more than the cautious default here. This simulator uses random draws from assumptions you control, and the defaults give a 4% rate an 82% result over 30 years. If you believe future returns will resemble the historical average, raise the return and see how the answer changes. The difference between the two approaches is the point: the rule of thumb hides how much depends on the assumption.

What happens in the year the money runs out?

The balance is set to zero and stays there. The scenario counts as a failure even if you have other income that could cover the gap.