Monte Carlo Simulation in Retirement Planning: What It Is and Why It Matters
Most free retirement calculators give you a number: “You’ll have $1.2M at 65.” This number assumes the stock market returns 7% every year, every year, in a straight line. It has never done that. The number is misleading.
Monte Carlo simulation is the alternative. Instead of assuming a fixed return, it generates thousands of different year-by-year return sequences — some great, some terrible, in random order — and asks: across all of these, in what percentage of scenarios does your plan survive to your target end age?
That percentage is the probability of success. It is a more honest answer to “will I be okay?” than any linear projection.
What Monte Carlo actually does
Step by step:
- The tool takes your inputs: portfolio value, annual withdrawal, asset allocation, retirement horizon (years)
- It generates a large number of random annual return sequences — typically 1,000 to 10,000 — statistically consistent with historical volatility (e.g., S&P 500 historical mean and standard deviation)
- For each sequence, it simulates your portfolio year by year: starting balance, market return, annual withdrawal, adjusted for inflation
- It checks whether each sequence ends with money remaining at the target end age, or runs out before it
- It reports the percentage of sequences where money remains: your probability of success
A 90% probability of success means 90% of the simulated sequences leave you with money at your target end age. The 10% that didn’t are scenarios where bad returns came early, compounding the damage of early withdrawals.
The key distinction from linear calculators
Linear calculators (Fidelity Retirement Score, Vanguard Nest Egg, most bank calculators):
- Assume a fixed 7% return every year
- Report: “Your projected balance at retirement is $X”
- Cannot model sequence risk (the damage of bad years early in retirement)
Monte Carlo tools (ProjectionLab, Boldin PlannerPlus, FIRECalc):
- Use randomised annual returns with realistic variance
- Report: “Your probability of success is X%”
- Do model sequence risk — this is the entire point
The practical difference: A plan that shows “you’re on track” in a linear calculator typically has 60-70% success probability in Monte Carlo. The same plan at 90% Monte Carlo success is a meaningfully more conservative plan. The linear calculator hid the gap.
This distinction is never disclosed on Fidelity’s or Vanguard’s free calculator pages. It’s why understanding Monte Carlo is not optional for serious retirement planning.
Run count matters
Not all Monte Carlo tools are equal. The number of simulated scenarios affects how accurately tail risks show up:
- 1,000 scenarios (Boldin PlannerPlus): Good; tail risks show up but with some noise at the extremes
- 10,000 scenarios (ProjectionLab): Better; tail risks are more statistically stable; better for long FIRE horizons
- Historical sequences (FIRECalc, cFIREsim): Different approach — uses actual historical data from 1871 onward; excellent for confirming survival through known worst-case sequences (Great Depression, 1970s stagflation, dot-com bust)
For conventional retirement (25-30 year horizon), the difference between 1,000 and 10,000 runs is modest. For FIRE horizons of 40-60 years, higher run counts provide meaningfully more reliable tail estimates.
What success rate to target
No universal rule, but common guidance:
| Success rate | What it means | Who it suits |
|---|---|---|
| Below 80% | Plan has meaningful risk of failure | Generally needs adjustment |
| 80-85% | Acceptable with significant spending flexibility | Early-career modeler with long runway |
| 85-90% | Solid plan with moderate flexibility | Most conventional pre-retirees |
| 90-95% | Conservative; good for those with no fallback | Pre-retirees without Social Security or pension backup |
| Above 95% | Often overly conservative — dying rich | Only if leaving a bequest is the explicit goal |
The right target depends on: whether you have Social Security (which reduces portfolio dependency), whether you have a pension (reduces portfolio dependency further), how flexible your spending is, and how long your horizon is.
Limitations of Monte Carlo
Monte Carlo is not perfect:
- It uses historical volatility, not future volatility. If volatility is structurally higher (or lower) in the future, the model’s assumptions change.
- It does not model correlation breaks. In crisis markets, asset correlations often move toward 1.0 (everything falls together). Monte Carlo built on normal historical correlations may underestimate crisis tail risk.
- It does not account for spending changes. Most Monte Carlo tools model a fixed or inflation-adjusted withdrawal, not the empirically observed “retirement spending smile” where spending declines by 1-2% per year in real terms after age 75. Conservative spending assumptions make the tool conservative.
- A 90% success rate is not a guarantee. It’s 90% of simulated sequences, which are themselves approximations of a future that hasn’t happened.
Despite these limitations, Monte Carlo is substantially more reliable than linear projection for the question “will I be okay?” The key is understanding what it measures and what it doesn’t.
Which tools to use
For most DIY planners, the practical answer:
- Start with FIRECalc (free): historical sequence testing since 1871. No registration required. Run your scenario and get a historical success rate.
- Cross-check with ProjectionLab free (Monte Carlo): independent second method. If both agree on roughly the same success rate, you have a convergent picture.
- For Roth conversion + tax optimisation: add Boldin PlannerPlus ($120/yr). The Monte Carlo is their tool’s weaker point (1,000 runs); it’s the tax modelling that justifies the price.
See our two-tool stack guide for the full workflow.
Sources: Bengen, W. P. (1994). “Determining withdrawal rates using historical data.” Journal of Financial Planning. IRS Publication 590-B. SSA.gov actuarial tables. CFPB retirement tools. Not financial advice.