Why "what's the average outcome" is the wrong question
If you only ever look at the average expected return, you're implicitly assuming the market delivers roughly that return every single year, in order. It never does. Two portfolios with the identical average annual return can produce wildly different results depending purely on when the good and bad years happen to land — a phenomenon called sequence-of-returns risk, and it's the single most important thing a Monte Carlo simulation shows you that a simple average never will.
Why this matters enormously more in retirement
While you're still adding money, a bad early year barely matters — you have decades to recover, and you're buying more units at lower prices the whole time. But once you switch to withdrawing money, a bad market in your first few retirement years can permanently damage the portfolio: you're forced to sell more shares at depressed prices to fund the same withdrawal, leaving fewer shares left to recover when the market eventually does. Two retirees with the exact same average lifetime return can have completely different outcomes purely because of which years were bad — this is exactly why the "probability money lasts the full period" number matters more than any single average-return calculation.
How to actually use this tool
- Don't just look at the median. Look at the 10th-percentile ("bad case") outcome and ask honestly whether you could live with it. A plan that looks great on average but fails badly in its worst 10% of outcomes is a fragile plan.
- Test your volatility assumption, not just your return assumption. Two plans with the same expected return but very different volatility (an all-equities portfolio vs. a diversified one) can show very different probabilities of success — this is the same trade-off our efficient frontier tool is built around.
- Re-run it periodically, not once. Your actual balance, contributions, and time horizon change every year — a plan is worth re-checking, not filing away.
The honest limitations
This tool assumes each year's return is independent and normally distributed, which real markets aren't quite — actual returns show fatter tails (more extreme years than a normal distribution predicts) and some clustering of good and bad years together. It also ignores fees, taxes, and inflation. Treat the probability number as a useful signal about robustness, not a precise, guaranteed odds calculation.