Monte Carlo Simulation for Retirement: Why Average Returns Lie
A retirement plan built on “6% average annual returns” looks fine on a spreadsheet and can fail badly in practice, because average returns don’t arrive in a straight line. Markets crash, recover, stagnate, and boom in unpredictable order. A 30% crash in your first year of retirement does far more damage than a 30% crash in year 20, even if the average return over the full period is the same.
Monte Carlo simulation runs your retirement plan through thousands of market scenarios with randomized returns and shows the probability that your money lasts. Instead of one answer (“you’ll be fine”), you get a distribution: “there’s an 89% chance your money lasts to age 95, a 7% chance it runs out between 85 and 95, and a 4% chance it runs out before 85.”
What Is Monte Carlo Simulation?
Monte Carlo simulation is a statistical technique that uses random sampling to model uncertainty. In retirement planning, it works like this:
- Define your plan: Current savings, annual contributions, spending target, retirement age, account types, tax rules, government benefits
- Set return assumptions: Expected return (e.g., 6%) and volatility (e.g., 15% standard deviation) for each asset class
- Run thousands of scenarios: Each scenario generates a random sequence of annual returns drawn from a probability distribution (typically log-normal)
- Track outcomes: For each scenario, simulate every year of retirement (withdrawals, taxes, benefits, account balances) and record whether the portfolio survives to the end
- Report the distribution: Success rate (% of scenarios where money lasts), percentile bands (5th, 25th, 50th, 75th, 95th), and worst-case outcomes
The name “Monte Carlo” comes from the casino: the technique relies on randomness, like a roulette wheel.
Why It Matters: Sequence of Returns Risk
The Problem With Averages
Consider two retirees who both experience 6% average returns over 30 years:
Retiree A: Gets +20% in years 1-5, then mixed returns after. Their portfolio grows early, giving a large base that sustains withdrawals easily.
Retiree B: Gets -15% in years 1-3, then strong returns after. Their portfolio drops early while they’re withdrawing, and the remaining balance never recovers, despite the same average return.
Both had “6% average returns.” One ran out of money. This is sequence of returns risk, and it’s the main reason Monte Carlo simulation is used in retirement planning.
The Math
A $1M portfolio with $40,000 withdrawn at the start of each year:
- Good sequence: +15%, +12%, +8% in years 1-3 → portfolio at about $1,244,000 after 3 years of withdrawals
- Bad sequence: -20%, -15%, +5% in years 1-3 → portfolio at about $608,000 after 3 years of withdrawals
Over 30 years, the two could still end up with the same average return if the bad-sequence retiree’s later years are stronger. The difference is timing: the bad-sequence retiree has about 51% less money and 27 more years of withdrawals ahead.
How to Read Monte Carlo Results
Success Rate
The headline number: “92% of scenarios survived to age 95.” This means that out of 1,000+ simulated market paths, 920 produced enough money to fund your spending through age 95. The remaining 80 ran out.
| Success Rate | Interpretation |
|---|---|
| 95%+ | Very comfortable: plan has significant margin |
| 85–95% | Good: manageable risk with some flexibility needed |
| 70–85% | Moderate risk: consider adjustments |
| Below 70% | High risk: plan likely needs changes |
Percentile Bands
The confidence bands show the range of possible outcomes:
- 95th percentile: your plan in a bull market; the estate could be very large
- 75th percentile: a better-than-average outcome
- 50th percentile (median): the typical outcome
- 25th percentile: below average; markets underperformed
- 5th percentile: near worst case; you need spending flexibility here
The gap between the 5th and 95th percentile is your uncertainty range. A wider gap means more outcome variability. The 5th percentile matters most for planning because it shows what happens if you’re unlucky.
What Monte Carlo Can and Cannot Tell You
What It Does Well
- Quantifies risk: it answers “how likely am I to run out of money?” with a probability.
- Tests strategies: Compare a Roth conversion ladder versus no conversions, or Guyton-Klinger spending versus fixed spending and see which produces better success rates.
- Reveals tail risk: The 5th percentile outcome shows what happens in the worst ~50 scenarios out of 1,000. If you can live with that outcome, the plan holds up.
What It Doesn’t Capture
- Fat tails: Real market crashes are more severe and clustered than the log-normal distribution assumes. 2008 was a 4+ standard deviation event, and Monte Carlo underestimates how often those happen.
- Behavioral risk: it assumes you stick to the plan. In practice people panic-sell in crashes and overspend in bull markets.
- Structural changes: it assumes the future looks statistically like the past. Shifts such as sustained low rates, demographic decline, or AI productivity gains aren’t modeled.
Monte Carlo shows probabilities; it doesn’t predict the future. Use it alongside historical backtesting (which uses actual market data going back to 1928) for a fuller picture.
How to Improve Your Success Rate
If your Monte Carlo success rate is below your comfort level:
- Reduce spending. The biggest lever. Even $2,000/year less in spending can improve success rates by 5–10 percentage points.
- Delay retirement. Each additional working year adds contributions and removes a withdrawal year. Even 1–2 years can shift the curve noticeably.
- Delay CPP/Social Security. Higher guaranteed income later means less reliance on the portfolio in your 70s and 80s.
- Use adaptive spending strategies. Guyton-Klinger guardrails or VPW allow spending cuts during downturns, which improves survival a lot.
- Optimize the withdrawal sequence. Drawing from the right accounts at the right time reduces tax drag and preserves tax-free growth.
- Reduce portfolio volatility. A higher bond allocation reduces return variability but also expected returns. The right mix depends on your risk tolerance and time horizon.
Monte Carlo vs. Historical Backtesting
| Approach | Method | Strength | Weakness |
|---|---|---|---|
| Monte Carlo | Random returns from a distribution | Tests thousands of paths, including ones that haven’t happened | Assumes returns follow a known distribution |
| Historical backtesting | Actual returns from 1928–present | Uses actual market data with real crashes and recoveries | Limited to nearly 100 years of data; future may differ |
Use both: Monte Carlo for a broad probability estimate and historical backtesting to ground it in real market behavior.
How cinder.fi Helps
cinder.fi runs 1,000+ Monte Carlo simulations in a Web Worker (keeping the UI responsive) with log-normal return distributions and configurable volatility. The Monte Carlo chart shows confidence bands from the 5th to the 95th percentile for your portfolio balance, spending, and estate value over time. The Summary tab shows the headline depletion probability alongside deterministic metrics. You can run any scenario, withdrawal strategy, or spending approach through Monte Carlo to compare outcomes. The US retirement calculator shows the success rate for your own plan.
Stress-test your retirement plan: try cinder.fi free.