Monte Carlo for Retirement: Running 10,000 Simulations to Test Your Plan
Your retirement plan has a 95% probability of success. But 5% of scenarios show you running out of money at age 82 instead of 95. Those 5% happen when the first 5 years have bad returns — sequence-of-returns risk. Here's how Monte Carlo simulation stress-tests retirement plans.
Monte Carlo simulation for retirement planning is a stochastic modeling technique that runs thousands of random return sequences to estimate the probability that a portfolio will sustain a given withdrawal rate over a specified time horizon. Unlike single-path deterministic projections (which assume constant returns every year), Monte Carlo acknowledges that markets are unpredictable and that the order of returns matters enormously for retirees. The technique generates 10,000 or more independent scenarios, each with a randomly generated annual return sequence based on the portfolio's assumed expected return and standard deviation. For each scenario, the simulator tracks the portfolio balance over time, subtracting annual withdrawals (typically adjusted for inflation). The percentage of scenarios where the portfolio survives to the end of the time horizon is the probability of success. A 90% success rate means the portfolio lasted in 9,000 out of 10,000 scenarios. Comprehensive retirement planning →
Why Monte Carlo is essential for retirement: Traditional retirement planning models assume a constant annual return — say 7% every year. This produces a smooth, predictable portfolio growth line. But real markets do not behave this way. A portfolio that earns -20% in year one and +30% in year two ends up very different from one that earns +30% in year one and -20% in year two, even though both have the same arithmetic average return. For someone withdrawing money each year, a bear market early in retirement can permanently damage the portfolio (a phenomenon called sequence-of-returns risk or sequence risk). Monte Carlo simulation captures this by randomizing the order of returns, showing the full distribution of possible outcomes rather than a single line. This allows retirees to understand not just the "most likely" outcome but also the tail risks — the scenarios where things go wrong. Sequence-of-returns risk →
Setting Up the Simulation
To run a Monte Carlo retirement simulation, you need six key inputs. The initial portfolio value is your current retirement savings. The annual withdrawal is how much you plan to take from the portfolio each year — typically expressed as a percentage of the initial portfolio (the withdrawal rate). The time horizon is how many years the portfolio needs to last, usually 25-30 years for a traditional retirement starting at age 65. The expected return is the annualized return you expect from the portfolio, based on its asset allocation (typically 7-9% for a 60/40 stocks/bonds portfolio). The standard deviation measures the volatility of returns (typically 10-15% for a 60/40 portfolio). The inflation rate is used to adjust withdrawals upward each year if you want constant purchasing power. Once these inputs are specified, the simulation generates annual returns using a normal distribution (or lognormal for prices), applies the withdrawal sequence, and tracks the portfolio balance. The simulation repeats this process 10,000 times, each with a different random return sequence, and calculates the success rate across all runs. Asset allocation by age →
Interpreting Success Rates and Percentiles
The output of a Monte Carlo retirement simulation is typically presented as a success probability: the percentage of scenarios where the portfolio did not run out of money. Financial planners generally consider an 80-90% success rate as acceptable, with more conservative planners targeting 95%. However, the success rate alone does not tell the full story. The simulation also provides percentile outcomes: the 10th percentile (worst 10% of scenarios), 25th percentile, 50th percentile (median), 75th percentile, and 90th percentile (best 10% of scenarios). The 10th percentile outcome is often the most important — it shows the portfolio value in the worst-case scenario that still occurs 10% of the time. If the 10th percentile portfolio drops to zero at age 82, there is a 10% chance of outliving your savings by age 82. The median outcome (50th percentile) shows what the typical scenario looks like — the portfolio might actually grow to $2M at age 95. The gap between the 10th and 90th percentile represents the range of uncertainty. A wide gap means more uncertainty in outcomes. Retirement income planning →
Withdrawal Rates and the 4% Rule
The single most important variable in retirement Monte Carlo simulations is the withdrawal rate. The famous 4% rule — withdraw 4% of the initial portfolio in your first retirement year, adjusting for inflation each year — was developed by William Bengen using historical data. Monte Carlo simulations broadly support this rule: a 60/40 portfolio with 4% inflation-adjusted withdrawals has approximately a 90-95% probability of lasting 30 years under reasonable assumptions. However, the rule is not a guarantee. Monte Carlo simulations show that withdrawing 5% instead of 4% drops the success rate to roughly 70-80%, and 6% drops it to 50-65%. The success rate is also sensitive to starting valuations — retiring when stock valuations are high (low expected future returns) requires a lower withdrawal rate. Many modern planners recommend 3-3.5% as a more conservative starting point, especially for early retirees facing a 40-50 year time horizon. Variable withdrawal strategies — where the withdrawal amount adjusts based on portfolio performance — can improve success rates by reducing withdrawals after poor returns. FIRE withdrawal strategies →
What is the difference between deterministic and Monte Carlo retirement projections?
A deterministic projection assumes a constant average return every year — for example, a 7% return every year for 30 years. This produces a single smooth projection line showing portfolio growth. It completely ignores sequence-of-returns risk. Monte Carlo simulation runs thousands of scenarios with randomized annual returns, producing a distribution of possible outcomes. Deterministic projections are misleading because they suggest a level of certainty that does not exist. Monte Carlo recognizes that the future is uncertain and provides a probability-based assessment. The deterministic approach will always show the portfolio surviving if the average return is positive; Monte Carlo shows that even with positive average returns, the portfolio can fail in some sequences.
How many simulations should I run for retirement planning?
Most retirement planning tools run 10,000 simulations, which provides approximately 1% accuracy in the success probability estimate. Running 100,000 simulations improves accuracy marginally to about 0.3%. For practical purposes, 10,000 is sufficient. The biggest source of error in Monte Carlo retirement planning is not the number of simulations but the quality of the input assumptions — the expected return, standard deviation, and inflation rate. Even a 0.5% error in the expected return assumption can change the success probability by 10-15 percentage points. Focus on getting the assumptions right rather than increasing the simulation count. The law of large numbers ensures that 10,000 simulations provide a stable estimate of the success probability.
What assumptions are most important for Monte Carlo retirement simulations?
The expected return assumption is the most critical and the most uncertain. A 60/40 portfolio assumed to return 8% will show a much higher success rate than one assumed to return 6%. The standard deviation (volatility) determines the dispersion of outcomes — higher volatility widens the range between best and worst cases. The withdrawal rate is the most actionable input. The time horizon is crucial: a 30-year retirement needs a lower withdrawal rate than a 25-year retirement. Inflation assumptions matter for inflation-adjusted withdrawals. Social Security, pensions, and other income sources must be modeled as separate cash flows for accuracy. The correlation between asset classes affects multi-asset simulations. Finally, the assumption about whether returns are normally distributed or have fat tails makes a significant difference in the tail risk estimates. Standard deviation and volatility →
How do I account for taxes in Monte Carlo retirement simulations?
Taxes add complexity to retirement simulations and are often the most challenging component to model accurately. Different account types (traditional IRA/401k vs Roth vs taxable) have different tax treatments. Traditional account withdrawals are taxed as ordinary income. Roth account withdrawals are tax-free. Taxable account withdrawals have capital gains taxes. A realistic Monte Carlo simulation should model the account-type-specific tax treatment, the tax bracket each withdrawal falls into, and the impact of Required Minimum Distributions (RMDs) from traditional accounts after age 73. Many retirees use a "taxable first, then tax-deferred, then Roth" withdrawal order to minimize taxes. The tax assumptions can significantly affect the success rate, especially for retirees with large traditional IRA balances. Some planners use a simplified tax rate assumption (e.g., 20% effective tax rate) while others build detailed tax bracket models. Tax-efficient fund placement →
Can Monte Carlo simulation model variable withdrawal strategies?
Yes. One of the most powerful applications of Monte Carlo simulation is testing variable withdrawal strategies. Instead of assuming a fixed inflation-adjusted withdrawal each year, the simulation can implement rules that reduce withdrawals after poor returns and increase them after strong returns. Common variable strategies include the guardrails approach (withdrawals increase or decrease by 10% when the portfolio changes by more than 20%), the percentage-of-portfolio method (withdraw a fixed percentage of the current portfolio value each year), and the Vanguard dynamic spending rule (withdraw previous year's amount adjusted for inflation but capped at ±5% of the portfolio percentage). Monte Carlo simulations consistently show that variable withdrawal strategies have higher success rates than fixed withdrawal strategies at equivalent initial withdrawal rates, because they reduce the damage from sequence-of-returns risk by cutting spending after bear markets. Systematic withdrawal plans →
Related Resources
Retirement Planning Guide
Build a retirement plan to test with Monte Carlo simulations.
Sequence-of-Returns Risk
The primary risk that Monte Carlo retirement simulation captures.
Standard Deviation and Variance
The volatility input that drives Monte Carlo dispersion.
Asset Allocation by Age
Choose the right asset allocation for your retirement time horizon.
FIRE Movement Guide
Early retirement withdrawal strategies tested by Monte Carlo.
Tax-Efficient Fund Placement
Tax-aware withdrawal sequencing in retirement simulations.