Interactive Calculator

#

How to Use This Calculator

#

Step 1: Choose Your Portfolio

#

Select from four professionally designed portfolios:

Returns are computed directly from each portfolio's asset assumptions and full covariance matrix using the portfolio-level Itô correction (μ minus half the portfolio variance). The calculator renders the same numbers live on each card.

Step 2: Set Your Withdrawal Rate

#

This determines how much you withdraw annually. The calculator supports two strategies:

Constant Dollar (Traditional): Fixed inflation-adjusted withdrawals regardless of portfolio value. Simple but rigid.

Dynamic Spending (Vanguard): Withdrawals adjust based on portfolio performance with floor/ceiling bounds. More sustainable for aggressive rates.

Quick guide:

• 2.5-3.0% = Very conservative (50-year horizons)
• 3.0-3.5% = Moderate (40-year horizons)
• 4.0%+ = Aggressive (30-year horizons or with flexibility)

Step 3: Set Duration

#

Enter years until age 100 (or your planning horizon). Add 5-10 years as a buffer.

Step 4: Add Fees (Optional)

#

Beyond ETF expense ratios, add any advisor or platform fees:

• 0.00% - DIY at Vanguard/Fidelity/Schwab
• 0.25% - Robo-advisors
• 1.00% - Traditional advisor (costs ~39% of wealth over 50 years!)

Step 5: Run Simulation

#

Click Run Simulation and wait 2-10 seconds. The default run is 10,000 scenarios. For tighter percentiles bump to 50,000 or 100,000 from Advanced Options (slower, more accurate).

Understanding Your Results

#

Depletion risk guidelines:

• 0-5% = Very safe
• 5-10% = Acceptable
• 10%+ = Consider reducing withdrawal rate

Advanced Options

#

Fat-Tail Mode: Layers a Student's t-distribution (df=5) on top of the default log-normal compounding to model extreme events (crashes and booms) more realistically than the bell curve alone. Recommended for conservative planning. If your plan survives fat-tail mode at a punishing withdrawal rate, your plan is genuinely robust.

Dynamic Spending Bounds: Adjust floor (-2.5% default) and ceiling (+5% default) to control withdrawal variability.

Privacy & Accuracy

#

All calculations run in your browser. No data is collected or stored.

The engine is professional-grade:

• Monthly Geometric Brownian Motion per asset. Returns compound via value × exp(r), so portfolios mathematically cannot fall below zero.
• Cholesky-decomposed correlation matrix drives correlated monthly shocks across the assets in the chosen portfolio.
• Fat-tail mode swaps the Normal driver for a variance-corrected Student's t-distribution (df=5).
• Annual rebalance at year-end. Withdrawals are taken at target weights (this has a monthly partial-rebalance side effect, disclosed in the methodology box inside the calculator).
• Sharpe and Sortino are computed per simulation path from monthly gross portfolio returns, then averaged across paths. That's the textbook formulation, not a cross-path outcome-dispersion proxy.
• S&P 500 comparison shares VTI's correlated monthly shock (empirical correlation 0.98). The "probability of beating the S&P" is therefore measured in the same market state, not in two parallel universes.
• Expected return labels on each portfolio card are derived directly from the asset assumptions plus the full covariance matrix, not hardcoded marketing numbers.
• Parallel no-withdrawal portfolio runs alongside the main simulation using the same Cholesky-correlated monthly shocks. Its median terminal value (and its full annual path on the chart) anchors the cost of your spending. If your median outcome is $16M and the no-withdrawal median is $40M, withdrawals cost you $24M of compound growth. The dashed brass line on the growth chart is that parallel portfolio's median path over time.

Open the Methodology and assumptions disclosure inside the calculator for the full set of caveats (no taxes, no cash buffer, historical correlations, the drawdown-includes-withdrawals convention).

The problem with average returns

#

If someone tells you stocks return 10% on average, your brain wants to multiply your portfolio by 1.10 every year for 30 years and put a number on the screen. That number is simply ridiculous.

Markets don't compound in straight lines. A simple example:

• Year 1: +20%. Your $100 becomes $120.
• Year 2: -10%. Your $120 becomes $108.

The arithmetic average of those two years is 5%. Your actual compound growth is 3.9%. Over 30 years that gap is the difference between retiring at 55 and retiring at 62.

Real markets are noisier than that. Some years are +25%. Some are -35%. The order matters. Monte Carlo simulation is the only way to capture both.

What Monte Carlo actually does

#

The name comes from the casino in Monaco, which is fitting because the whole approach is built on probability.

Here is the process:

1. Define your inputs: starting balance, withdrawal rate, time horizon, asset allocation.
2. Generate a random sequence of monthly returns using each asset's historical mean and volatility.
3. Compound through the horizon. Withdraw money each year. Track the path.
4. Repeat thousands of times. Look at the distribution: how many runs survived, how many depleted, what the median outcome looked like.

Each run is one possible future. Maybe you retire and immediately eat a crash. Maybe you get a lucky decade of bull market right at the start. Monte Carlo shows you the whole range so you can plan for the bad ones, not just hope for the good ones.

Sequence of returns risk is the real fight

#

This is the part that keeps early retirees awake at night.

While you are still working and adding money, a crash early in your career is great. You buy cheap. Compounding loves you back.

After you stop working and start withdrawing, an early crash is a different beast entirely. You are selling assets to fund your life right as those assets get cheap. Every dollar you pull out at the bottom is gone. The portfolio that should have recovered now has less left to recover with.

Two retirees, same long-run average return, different sequence:

Same average. Five times the wealth gap. That is sequence of returns risk in one table, and it's exactly what Monte Carlo captures by randomizing the order of returns across thousands of paths.

The 4% rule and where it stops being safe

#

You've probably heard the 4% rule: withdraw 4% of your starting balance each year, adjust for inflation, and a 30 year retirement should hold up. It comes from the Trinity study, which used US market data from 1926 to 1992.

That is fine for a traditional retirement at 65. But it starts breaking when you stretch it.

What the simulations consistently show:

Those aren't guarantees. They are probability distributions. The point of running thousands of scenarios is to stop asking "will it work" and start asking "in what fraction of futures does it work, and am I OK with the rest."

Fat tails and the fix I had to make

#

Most Monte Carlo tools, including the first version of mine, generate returns from a Normal distribution. Bell curve. Smooth. Reassuring. The problem is that real equity markets have fat tails. Extreme events happen far more often than the bell curve predicts.

Just since 2000:

• 2000 to 2002: dot-com bust, Nasdaq down 78%.
• 2008: global financial crisis, S&P 500 down 57%.
• 2020: COVID crash, down 34% in three weeks.

A Normal distribution says crashes that severe should be once-per-century events. We've had three this century, and we're not done. Not that long ago we had liberation day and now the Iran War.

There's a second, deeper problem with Normal returns: arithmetically, a Normal return r and the compounding step value × (1 + r) can drive a portfolio below zero. That is mathematically impossible (a stock can't be worth less than nothing) but a naive simulator will happily print it.

The fix is to model returns as log-normal, which is what Geometric Brownian Motion actually does. Instead of value × (1 + r), you compound with value × exp(r) where r is a normal log-return. Log-normal can't go below zero by construction, naturally produces a heavier right tail, and matches the empirical distribution of monthly equity returns much better than Normal does.

My calculator now uses log-normal compounding for the default mode. The fat-tail toggle layers a Student's t-distribution (df=5) on top, which gives even heavier tails for stress testing. If your plan survives the fat-tail mode at a punishing withdrawal rate, your plan is genuinely robust. If it doesn't, you've found the edge of your plan before reality finds it for you.

Picking a withdrawal strategy

#

Monte Carlo lets you stress test the spending side too.

Constant dollar. You withdraw a fixed inflation-adjusted amount every year regardless of what the portfolio is doing. Simple. Predictable. You will be pulling the same dollar amount out of a declining portfolio during a crash, which is exactly when you shouldn't.

Dynamic spending (Vanguard rule). You adjust withdrawals based on portfolio performance with floors and ceilings so spending doesn't swing wildly. That's what I'm using. More sustainable at aggressive withdrawal rates. The trade-off is that you have to be willing to cut spending in bad years.

The simulations are consistent: at the same nominal withdrawal rate, dynamic spending often cuts depletion risk roughly in half.

What the numbers cannot tell you

#

Monte Carlo is powerful. It also has blind spots.

Regime change. The simulation assumes future volatility looks like past volatility. What if we enter a Japan-style decade of low returns? The simulator doesn't know.

Structural shifts. AI rewriting the labor market. Demographics. Climate. None of this is in the model.

Personal factors. Health expenses, family obligations, a roof that needs replacing. The simulator doesn't know your life.

Taxes. Most simulators work in nominal returns. Your actual spending power depends on your tax situation, which for a Dubai-based expat looks very different from someone in London or Paris.

Use Monte Carlo as one input. Don't treat the 50th percentile as a forecast. Treat the 5th percentile as a planning floor.

Build in safety margins on top

#

Don't just trust the simulator's 5th percentile result.

• Regime change buffer: -20%. Persistent low returns are a real possibility.
• Black swan buffer: -15%. Major crisis early in retirement.
• Fee creep buffer: -5%. Costs tend to climb over decades.

If the 5th percentile outcome in the simulator says you have $1.7M, your conservative planning target after those haircuts is closer to $1.0M. If you can live on that conservatively-haircut number, you are genuinely safe. Not "statistically probably fine." Safe.

Try it yourself

#

Theory is one thing. Running your actual numbers through the simulator is where it stops being abstract.

Test different withdrawal rates. Compare portfolios. Watch how fees compound over decades. Everything runs in your browser. No accounts, no tracking, no data leaving your machine.

The bottom line

#

Simple retirement calculators sell false precision. They give you one number and pretend to know the future. Monte Carlo is honest about uncertainty. It hands you a distribution and lets you decide how much downside you're willing to plan for.

You might not love seeing a 15% chance of running out of money. You would love it less at 85.

Plan for the 5th percentile. Hope for the median. Stay flexible enough to adjust when reality surprises you, because it will.