Measure
Monte Carlo simulation
Estimating a distribution of outcomes by simulating many return paths.
Definition
Estimating a distribution of outcomes by simulating many return paths.
How to read it
It replaces a single historical sequence with thousands of possible ones, showing the spread of outcomes rather than one answer.
Measured on real funds
510000 simulated 30-year retirements, drawn by 12-month blocks from VTI's own 303 months of history rather than from a bell curve. $1,000,000 start, $40,000 a year withdrawn.
| Outcome | Ending balance | Meaning |
|---|---|---|
| Worst 5% | 726,083 | 1 run in 20 ended at or below this |
| Worst 25% | 4,809,767 | 1 run in 4 |
| Median | 10,852,509 | half above, half below |
| Best 25% | 22,005,964 | 1 run in 4 did better |
| Ran out of money | 2.4 | percent of runs reaching zero |
Block bootstrap keeps the runs and the fat tails that real markets have and a normal distribution removes. Past sequences are not a forecast; they are the only sample there is. Computed from Yahoo Finance monthly total return, recomputed with every update of the site. How these figures are produced. Run it on your own numbers — free, no account.
What it does not tell you
The output is a restatement of the input assumptions. Draws from a normal distribution will understate crashes no matter how many paths are run, and a precise-looking success rate expresses confidence in a model rather than in the world.
Applied in 5 articles
5- How Long Will Your Money Last? The 4% Rule Tested on 25 Years of Real Fund Returns
- Monte Carlo vs the 4% Rule: Simulating Retirement Withdrawals Instead of Assuming Them
- The Master Class: How to Use Claude (Anthropic) to Backtest Your Own ETF Strategy
- "The Curse of Sequence Risk" – Protecting Your 30-Year Plan from a Pre-Retirement Crash
- Re-evaluating the 4% Rule: Sequence Risk, Yield, and Dynamic Withdrawal in 2026
Every figure computed with this measure on this site follows the conventions set out in the methodology. All measures →