Monte Carlo Simulation for Retirement Income: Will It Last?

3.9%. Morningstar found that this starting withdrawal rate still had money left in 90% of simulated futures over a 30-year retirement (Morningstar, The State of Retirement Income 2025, Dec 2025). On $1 million, that's $39,000 in year one. Researchers get numbers like that from a Monte Carlo simulation for retirement income, and you can use the same method to test your own plan. It runs your savings and spending through many possible market futures and counts how many stay funded. What you get back is a probability. Not a promise.

Below: how the method works, how to read your success rate, where the research disagrees, and what to check before you trust any calculator.

What a Monte Carlo simulation for retirement income actually does

Picture retiring 1,000 times.

In one of those retirements, the market crashes in year two. In another, it booms for a decade. In a third, inflation flares just as you start withdrawing. The simulation picks plausible returns for each year, takes out your spending, and checks whether any money is left at the end. Then it counts the survivors. If 900 of 1,000 paths stay funded, your success rate is 90%.

Why bother, when you could just assume an average return? Because nobody retires into the average. You retire into one particular sequence of returns. Take two retirees with the same average return. If one gets the bad years early, while they're withdrawing from a full balance, they can end up in very different places. That's sequence-of-returns risk, and an average-return projection can't see it (Kitces).

Typical inputs include (T. Rowe Price):

• Current balance and contributions
• Spending, or withdrawals
• Asset allocation
• Expected returns and volatility (how much returns swing from year to year)
• Inflation
• Time horizon

How to read your success rate

This is where most people go wrong.

A 90% success rate doesn't mean a 10% chance of going broke. As White Coat Investor explains, it means roughly a 10% chance you'll need to make adjustments along the way. Real retirees spend a bit less in bad years. Most simple models assume you never do.

Two more cautions:

• Fixed spending isn't real behavior. Many models assume the same inflation-adjusted withdrawal every year. They often leave out pensions or rental income too. People shown a single score also tend to underspend (ThinkAdvisor).
• The score doesn't tell you how big a failure is. A path that runs short by a little at 94 counts the same as one that empties at 78 (Kitces).

So read the number as a measure of how fragile your plan is. It isn't a countdown.

A worked example in dollars

Say you've saved $800,000. Here's what two research-backed starting rates come to in year one:

The 3.9% figure is Morningstar's 2025 base case, up from 3.7% the year before. It leaves out Social Security, so any benefits come on top of it (Morningstar). Morningstar also found that flexible spending rules can push the starting rate to about 5.7% (Morningstar Australia).

That's a $14,400-a-year gap. It shows what being willing to adjust can be worth. Your own number depends on your timeline, your mix of investments and your other income. That's why the generic rule is where you start, not where you finish.

Where Monte Carlo can mislead you

The assumptions decide the answer

The results are "entirely determined by the CMAs used" (Advisor Perspectives). CMAs are capital market assumptions: the future returns and volatility the model plugs in. If a tool assumes rosy returns, it gives you a rosy success rate. Same engine, different inputs, different verdict.

Monte Carlo vs. history: the research disagrees

Researchers have also compared Monte Carlo with historical simulation, which replays actual past market sequences. The findings point in different directions:

• An earlier analysis found that a 93.5% Monte Carlo success rate roughly matched a 100% historical success rate. In other words, Monte Carlo looked more cautious (Kitces).
• Later work found that at moderate risk levels, Monte Carlo estimated about 5–10% more sustainable income than historical data (Kitces).
• At the 70–90% success rates advisors commonly use, Monte Carlo can understate risk compared with history (Income Lab).
• A follow-up study by Fitzpatrick and Tharp found traditional Monte Carlo was the most error-prone of the methods tested. Historical simulation was more accurate at common success levels, and regime-based Monte Carlo, which models markets moving through distinct good and bad periods, was more conservative (Income Lab).

How can both be true? Our reading is that they're looking at different parts of the curve. Near 100%, you're asking about the very worst paths. Random simulation can invent disasters worse than anything in the historical record, so it looks cautious there. In the 70–90% band, you're asking about ordinary bad decades. Traditional models treat each year as independent, so they may smooth over the way rough stretches tend to bunch together in real markets.

The practical lesson: don't treat 85% versus 88% as precise. Read the score as a range, and expect the model behind it to shift it a few points either way.

Ask how late, not just how often

A good tool should tell you something about the failing paths. At what age does the money run short? By how much? Hitting a shortfall late can be handled with small changes. Hitting one early is a different problem.

Three risks your simulation has to include

Longevity. About one in three 65-year-olds may live to at least 90 (Social Security Administration, Retirement Information for Medicare Beneficiaries, 2026). A plan that stops at 85 quietly bets against a third of retirees. That's why plans built in Monte Climb run to 95 by default, and you can push them further.

Health care. Fidelity estimates that a 65-year-old retiring in 2025 will need $172,500 for Medicare premiums, deductibles and prescriptions over retirement. That figure leaves out long-term care entirely (Fidelity Investments, 2025 Retiree Health Care Cost Estimate, Jul 2025). It's a real line item, so it should be modeled year by year, including income-based Medicare surcharges, not covered by one flat guess.

Sequence risk. Covered above, and it bears repeating: the first years of retirement carry the most weight.

A checklist for judging any retirement calculator

Before you trust a success rate, ask:

1. What return and volatility does it assume? Can you see them?
2. Is inflation modeled? At what rate?
3. Does it count Social Security, pensions or other income?
4. How far does the plan run? Is 85 the cutoff, or 95 and beyond?
5. Is health care its own line item? Or is it buried in "spending"?
6. Can spending flex? Guardrail rules, such as Guyton-Klinger, cut spending when the portfolio falls past a set threshold and raise it when markets do well (ThinkAdvisor).
7. Does it show how bad and how late failures are? Or only a percentage?
8. Does it show its work? If you can't see why the number moved, you can't act on it.

Start with three numbers

Here's a striking one: 48% of workers have never estimated how much income they'll need each month in retirement (EBRI/Greenwald, Retirement Confidence Survey 2025, Fact Sheet 3). Half are guessing.

You don't have to. Enter your age, what you've saved, and what you spend each month, and Monte Climb's retirement calculator runs those numbers through 1,000 simulated futures and returns a success rate. It runs in your browser, and nothing is saved until you say so. Then ask William, our AI planner, the follow-up questions: Am I on track? Can I retire at 62? What's a sustainable spend? He shows his work.

It's planning software, not advice. But it's a lot better than guessing.

Show my answer →

Frequently asked questions

What is a Monte Carlo simulation for retirement income?

It's a way to test a retirement plan against many possible market futures, often 1,000 of them. Each run uses a different sequence of plausible returns, takes out your spending, and checks whether money is left at the end. The share of runs that stay funded is your success rate.

Does a 90% success rate mean a 10% chance of running out of money?

Not exactly. It means that in about 10% of simulated futures, the plan as written doesn't hold up, so you'd probably need to adjust along the way, for example by spending less in bad years. Most simple models assume you never adjust, which real retirees usually do.

What is a good Monte Carlo success rate for retirement?

There's no single right target. Morningstar's 2025 research uses a 90% success rate over a 30-year retirement as its benchmark, which gives a 3.9% safe starting withdrawal rate. The research also shows results can move by several points depending on the model and its assumptions, so read your score as a range rather than an exact figure.

Is Monte Carlo more accurate than historical simulation?

The research is mixed. Some studies found Monte Carlo more conservative than history at very high success rates. Others found it can understate risk in the 70–90% range many plans fall into. A study by Fitzpatrick and Tharp found traditional Monte Carlo the most error-prone of the methods tested, while regime-based Monte Carlo was more conservative.

What should a retirement Monte Carlo simulation include?

Look for visible return, volatility and inflation assumptions, Social Security and other income, a time horizon that runs well past 85 (longevity), health care costs modeled separately, the option to let spending flex, and some detail on how severe and how late any shortfalls are.