FIRE Dashboard

Monte Carlo simulation: how likely is it that your wealth lasts?

Nobody knows how the stock market will do over the next thirty years. A Monte Carlo simulation does not pretend to know. It plays out your plan a thousand times, each time with different good and bad years, and counts how often your wealth lasts. So you do not see one outcome, but how wide the range of possible outcomes is.

In short

  • A single average return hides the biggest risk: the order in which the good and bad years come.
  • A Monte Carlo simulation plays out your plan thousands of times with random years, takes out your spending every year and counts how often there is still wealth left.
  • A 90% success rate means that in 9 out of 10 simulated futures your wealth did not run out. It is not a guarantee.
  • Look mainly at the bad outcomes, such as the tenth percentile, and keep a margin.
The Monte Carlo simulation in FIRE Dashboard, in just over a minute. With example figures; the voice is AI-generated.

Why one average return is misleading

A simple FIRE calculation uses a fixed return, say 6% a year. But the stock market never gives 6% every year. One year it is +25%, another −20%. As long as you are only saving and investing, the order of those years makes little difference. Once you take money out every year, it makes a big difference.

An example. Two people stop working with €1 million each. Both take out €40,000 at the start of every year. They get the same five returns, 4% a year on average, only in reverse order. Without withdrawals, both would end up with €1,138,500. With withdrawals, they do not:

YearReturn AWealth AReturn BWealth B
1−20%€768,000+25%€1,200,000
2−10%€655,200+15%€1,334,000
3+10%€676,720+10%€1,423,400
4+15%€732,228−10%€1,245,060
5+25%€865,285−20%€964,048

After five years, B has almost €100,000 more than A, with the same average. In the bad years, A sold investments at low prices to live on. Those euros are gone and cannot recover when the market does. This is called sequence-of-returns risk: bad years just after your FIRE date do far more damage than the same bad years later on. A calculation with one average return cannot see that risk.

Try it with your own numbers, free

What a Monte Carlo simulation does

The name comes from the casino in Monaco: the simulation rolls the dice again and again. This is how it works:

  1. Expectation and spread. For your investments, you choose an expected return per year and how much it swings (the volatility). Shares, for example, have a spread of around 15% a year; savings hardly any.
  2. Random years. For every year, the computer draws a return that is sometimes better and sometimes worse than expected. On average it comes out at the expected return.
  3. Withdrawals. Every year your spending is taken out, raised with inflation, just as in real life.
  4. Repeat. The computer does this a thousand times or more. Each run is a possible future, with its own order of good and bad years.
  5. Count. In how many of those futures is any of your wealth left at the end? That is the success rate.

How to read the result

The success rate is the share of futures in which your wealth lasts until the end of the period, for example 30 or 40 years. If that is 900 out of 1,000, the success rate is 90%.

The percentiles show how wide the outcomes are. Line up all the futures, from bad to good:

  • The tenth percentile (P10): in 10% of the futures you end lower than this amount. This is a bad outcome, but not the very worst.
  • The median (P50): the middle outcome. Half end lower, half end higher.
  • The ninetieth percentile (P90): in 10% of the futures you end higher than this amount.
€0 €1m €2m €3m €4m 0 10 20 30 years P90 median P10
between P10 and P90 · median. Example: €1 million, taking out €40,000 a year and 2% more each year, fully in shares with an average return of 7% and a spread of 15%, without tax. In today's euros. Success rate after 30 years: 75%.

The fan gets wider and wider: the further away, the less certain. And it is lopsided. The good outcomes run far up, while the bad ones stop at zero. In the example, in 10% of the futures your wealth has run out after about 22 years, while the median after 30 years is still at about €650,000.

What 90% does and does not mean

  • It does mean that in the model, with the assumptions you entered, your wealth lasts in 9 out of 10 futures.
  • It does not mean that the real-world chance is 90%. The result is only as good as the assumptions. A slightly lower expected return or a bigger spread changes the success rate a lot.
  • It does not mean that in 1 out of 10 cases things go wrong the way they do in the model. In the model you keep spending the same no matter what. In real life you would probably spend a bit less after a few bad years, or do some work for a while.
  • It does not tell you how well things turn out. A future that ends with €1 left counts as a success just as much as one that ends with €3 million. So look at the percentiles as well.

What the simulation in FIRE Dashboard does

The simulation in FIRE Dashboard starts with your FIRE number at your FIRE age and looks 30, 40 and 50 years ahead. It is part of the free version. What it takes into account:

  • Each asset class separately. Savings, shares, property, crypto and other assets such as gold and silver each have their own expected return and their own spread. A portfolio of shares and savings therefore moves more realistically than with one average return for everything.
  • A shared market factor. The asset classes partly move together, as they do in real life. You set how strongly with the correlation, 50% by default. If you'd rather not tune individual numbers, pick one of the presets Calm, Normal or Rough; the individual settings are under Advanced. At 0% they move independently and diversification looks too good; at 100% everything moves in lockstep.
  • Crash years. If you switch this on (it is on by default), the market shock in about one year in ten is much larger than normal, as in 2008. Large falls then happen more often than with a normal distribution. The expected return stays the same.
  • Yearly rebalancing. If you switch this on (it is on by default), your portfolio goes back to the mix you started with every year. Without rebalancing, the part with the best returns keeps growing, and so does the risk.
  • Box 3 every year. After each year, the box 3 tax is taken off your personal wealth.
  • AOW and pension. From the year they start, they reduce what you need to take out of your wealth.
  • Spending with inflation. Your withdrawal rises every year with the inflation you set. You can also see the final amounts in today's euros.

For each horizon you see the success rate, the median and P10 and P90. You choose how many futures the simulation plays out; the default is a thousand. If you have a BV, the simulation uses two pots: personal wealth with box 3, and the BV, where every withdrawal pays corporate income tax and box 2. How that works is explained in Monte Carlo with a BV.

The stress tests next to it

A simulation shows how often something goes wrong. A stress test shows what happens if one particular bad scenario comes true. FIRE Dashboard has four, also free, next to a base path with the expected return every year:

  • A crash in the first year after your FIRE date: exactly the sequence-of-returns risk from the example above.
  • The same crash after five years: so you see that the timing of the blow matters.
  • High inflation: a number of years in which your spending rises much faster than normal.
  • A lost decade: ten years in which your investments earn little.

If your plan holds up in all four, you have a solid base. If it fails in one, you know what to watch out for.

What a simulation cannot do

  • It is still a model. The future is unknown. The simulation draws years from a distribution you choose; the real market does not follow it.
  • The assumptions drive the result. An expected return of 7% instead of 6% makes a big difference. So also run it with cautious assumptions.
  • Rules change. The simulation uses today's tax rules. Over thirty years, those can change many times.
  • It is not advice. FIRE Dashboard is a calculation aid. For big decisions, it is best to consult an adviser.

Practical tips

  • Keep a margin. Many people aim for a success rate of 85 to 95%. Above 95% often means working longer than you need to; below that means you need to be able to adapt.
  • Look at the tenth percentile. Could you live with that outcome? Then you are in good shape. If not, a lower withdrawal rate or a bigger buffer helps.
  • Plan for flexible spending. Spending 10% less for a while after a bad year reduces sequence-of-returns risk considerably. Decide in advance where you could cut back.
  • Watch the first years. A cash buffer for a few years of spending means you do not have to sell investments at low prices after a crash.
  • Compare scenarios. Put a cautious and a normal version side by side. The difference shows how sensitive your plan is.

Sources

  • William P. Bengen, Determining Withdrawal Rates Using Historical Data, Journal of Financial Planning, 1994
  • Cooley, Hubbard and Walz, Retirement Savings: Choosing a Withdrawal Rate That Is Sustainable, AAII Journal, 1998

A calculation aid, not financial advice. The example and the chart are an illustration, not a forecast. Past returns are no guarantee for the future.