Coast FIRE Monte Carlo Calculator

How the simulation works

The regular Coast FIRE calculator gives your portfolio the same average return every year. This one gives it 2,000 different futures and counts how many of them work out.

FIRE number = yearly spending ÷ withdrawal rate
real return = (1 + expected return) ÷ (1 + inflation) − 1

For each of 2,000 paths, for each year until retirement:
  growth  = exp( ln(1 + real return) + volatility × Z )    Z is a fresh standard normal draw
  balance = balance × growth + 12 × monthly contribution   (only while younger than your stop age)

chance of success = share of paths where balance at retirement ≥ FIRE number

Each year’s return is drawn at random from a lognormal distribution. Its median is your expected real return, and its spread is set by the volatility you choose, which is the yearly standard deviation of log returns. A lognormal can’t go below −100%, which is one reason it’s a common choice for this kind of model. Every year is drawn independently of the last.

Contributions go in once a year and stop at the age you pick. The default stop age is the Coast FIRE age the regular calculator gives for the default inputs, rounded up to a whole year. Once you change your inputs, the table below uses your own coast age. The result is the share of the 2,000 paths that end at or above your FIRE number on your retirement date.

The chart shows three lines through all those paths: the 10th percentile (a bad but not disastrous outcome), the median, and the 90th percentile. The table underneath reruns the whole simulation as if you stopped at the first whole age after your coast age, three years later and five years later. The simulation uses a fixed random seed, so the same inputs always give the same answer.

A worked example

Maya is 30, has $80,000 invested, adds $2,000 a month and wants to retire at 65 on $60,000 a year. Her FIRE number is $1,500,000, and the regular calculator says she reaches Coast FIRE at 48. The simulation works in whole years, so the first age she can stop at after that is 49. Here’s how often she still reaches $1,500,000 by 65, depending on when she stops contributing:

Both columns use the same 7% expected return, to keep the comparison simple. In real life a balanced portfolio would usually get a lower return as well.

Stopping at 49 works about half the time. That surprises people, but it follows from how the regular calculator works: its Coast FIRE age is the point where average returns get you exactly to your number, and in a simulation built around those same average returns, the typical path lands right on it. In the stock-heavy case, Maya’s median balance at 65 is $1,559,657, while her 10th percentile is $612,747.

The more useful lesson is how slowly the odds improve. Five more years of contributions only move her from 52% to 60%. Even contributing every month until 65 leaves her at 73%. At 49 her balance on paper is about $817,021, big enough that $24,000 a year of new money barely changes where it ends up. What decides the outcome is 35 years of returns, and no contribution schedule can fully insure against those.

So I wouldn’t read a low number as a reason to never stop. I’d read it as a reason to keep some flexibility: a willingness to work a couple of extra years, or to restart contributions if the first decade after you stop goes badly.

How do I choose the volatility?

Volatility is how much returns swing from year to year. A higher number widens the spread of outcomes, pushing the 10th percentile down and the 90th up.

The 15% default fits a portfolio that’s mostly stocks. For context, S&P 500 total returns have had a yearly standard deviation of roughly 19% since 1928, by my count from Aswath Damodaran’s historical returns data at NYU Stern. That’s for an all-stock index, and the calculator’s measure is slightly different (it uses log returns), so treat 15% as a middle-of-the-road setting rather than a historical fact. A mix closer to 60% stocks and 40% bonds swings less. On the same data it comes out nearer 12%, so 10% is a slightly gentle setting for it.

If you want a stress test, try 18% or 20% and see whether your plan still looks acceptable.

What this model doesn’t capture

It’s a simple model, and it errs in predictable directions.

Real markets have fatter tails than a lognormal distribution, so severe crashes show up more often in history than they do in this model. Years also aren’t independent in real life. Long stretches of poor real returns, like the high-inflation 1970s, are a kind of risk this simulation only produces by chance.

Inflation is fixed at whatever you enter, when in reality it moves around from year to year. Fees and taxes aren’t modeled, so subtract fees from your expected return. Your contributions are assumed to arrive every year until your stop age, with no job losses or career breaks.

Finally, it only looks at the years before retirement. Sequence risk after you retire, where a bad market in your first few years of withdrawals does lasting damage, is outside this model. That’s what the withdrawal rate is meant to cover, and why a lower one buys real safety.