Your financial planner ran 10,000 simulations and told you there is a 92% probability of success. That number is not a measurement. It is a function of which historical window was sampled, which return assumptions were chosen, and which definition of "success" was used. Change any one of those inputs and the number changes by 20 points. Pension funds do not use Monte Carlo to decide whether they are funded. Neither should you.

The Ritual

You sit down with a financial planner — or you open cFIREsim, or FireCalc, or Boldin, or any of two dozen tools the FIRE community trusts. You enter your portfolio, your spending, your age. The software runs 10,000 simulations. Each simulation draws a sequence of returns from historical data (or from a statistical distribution fitted to historical data), applies your withdrawal rate, and checks whether the portfolio survives 30 years.

The output: "92% probability of success."

It feels rigorous. 10,000 trials. Each one different. A probability, not a guess. You feel measured. You feel modelled.

You are not. You are watching a performance.

What Monte Carlo Actually Does

A Monte Carlo simulation samples returns from a distribution. The question is: which distribution?

If the simulation draws from US historical returns (1926–2025), you are implicitly assuming that the next 30 years will be statistically similar to the last 100. This is not a neutral assumption. The US equity market in the 20th century was the most successful in human history. Sampling from it is sampling from the winner. An investor in Japan in 1989 — identical portfolio, identical spending — would get a very different "probability of success."

If the simulation draws from a parametric distribution (mean 7%, standard deviation 15%), you are assuming returns are independently and identically distributed — that each year's return is statistically independent of the last. This is false. Returns are serially correlated. Real interest rates are persistent. Inflation regimes last decades. The distribution itself changes over time — and it changes most violently in exactly the states of the world where your retirement is most at risk.

The output of a Monte Carlo simulation is a function of the assumptions. It is not a measurement of anything about you.

Change the historical window from 1926–2025 to 1966–1995 (which includes the inflationary 1970s and early 1980s) and your 92% drops to 74%. Change the expected return assumption from 7% to 5% and it drops to 61%. Change "success" from "portfolio survives 30 years" to "portfolio maintains purchasing power for 40 years" and it drops further.

None of these changes reflect anything that happened in your portfolio. They are modelling choices made by the person who built the tool. The "probability" is not a property of your retirement. It is a property of the simulation.

The Deeper Problem

Even if Monte Carlo could perfectly sample future return distributions — which it cannot — it would still answer the wrong question.

"What is the probability that my portfolio survives 30 years?" is a question about terminal wealth. Will the pot of money last? Will I run out?

That is not the question you should be asking. The question you should be asking is: "What standard of living can my assets sustain, with certainty, for as long as I live?"

These are fundamentally different objectives. The first optimises for the balance not reaching zero. The second optimises for consumption — a stable, inflation-protected stream of income that lasts a lifetime. The first treats retirement as a wealth problem. The second treats it as an income problem.

The entire personal finance industry is built on the first question. The entire pension fund industry is built on the second. The pension funds are right.

Here is why. A 60-year-old with $1,000,000 who runs a Monte Carlo simulation and gets "95% probability of success at 4% withdrawal" knows one thing: in most of the simulated histories, the money lasted 30 years. What she does not know:

  • Whether her spending is hedged against inflation — or whether 3% sustained inflation over 20 years will erode her purchasing power by 45%.
  • Whether her portfolio's duration matches her liability's duration — or whether a fall in real interest rates will make her future spending more expensive in present-value terms while her portfolio barely moves.
  • Whether her assets carry good beta or bad beta — whether a market crash will be self-correcting (discount-rate news) or permanent (cash-flow news).
  • How much of her consumption is already funded by Social Security — an asset the simulation ignores entirely.

The Monte Carlo says "95%." It does not say 95% of what, against which risks, with what hedged and what exposed. It is a single number that obscures every dimension of risk that actually matters.

What Pension Funds Do Instead

CalPERS does not run a Monte Carlo simulation and report "93% probability of meeting obligations." What CalPERS does — what every defined-benefit fund does — is construct a balance sheet:

Left side: assets. Valued at market.

Right side: liabilities. The present value of every future payment owed, discounted at the real risk-free rate, weighted by the probability that the beneficiary is alive to receive it.

The ratio: funded ratio = assets ÷ liabilities.

If the funded ratio is 0.85, CalPERS knows: we have 85 cents for every dollar promised. The gap is 15 cents. And CalPERS knows what kind of risk is driving the gap — because the liability has measurable characteristics: duration, inflation sensitivity, real-rate sensitivity. The assets can be compared to the liability on every one of those dimensions.

This is not a simulation. It is a measurement. And the measurement tells you exactly what to do about it: match the duration. Hedge the inflation exposure. Close the gap with assets whose character matches the character of the obligation.

You have the same balance sheet problem. Your retirement is a stream of real spending, stretching decades into the future, with known duration and known inflation sensitivity. The question is not "will my money last?" The question is: "Do my assets match my obligations?"

Monte Carlo cannot answer that question. It does not even ask it.

"But My Monte Carlo Is Sophisticated"

The sophisticated objection: "I don't use simple historical bootstrapping. I use regime-switching models, or stochastic volatility, or correlated return-and-inflation draws. My Monte Carlo is better."

The strongest version of this objection comes from Karsten Jeske — Big ERN — whose Safe Withdrawal Rate series on Early Retirement Now is the most rigorous analysis in the FIRE community. His CAPE-adjusted return forecasts and dynamic withdrawal rules represent a genuine improvement over raw historical bootstrapping. I recommend it to anyone who wants to understand sequence-of-returns risk in depth. But even ERN's framework shares the structural limitation: it optimises for portfolio survival — the balance does not reach zero — rather than consumption certainty. Two portfolios can both show "95% success" in ERN's framework while having wildly different funded ratios, because one has hedged 80% of its consumption liability with TIPS and the other has hedged nothing. The SWR series asks "does the portfolio survive?" The funded ratio asks "is the spending funded?" These are different questions with different answers, and the second one is the one your retirement depends on.

Perhaps. But even a perfectly specified Monte Carlo shares a structural limitation: it produces a probability of portfolio survival, when what you need is a diagnosis of hedging gaps.

Consider two investors. Both have $1,000,000. Both spend $40,000/year. Both get "90% probability of success" from the same Monte Carlo engine.

Investor A holds $800,000 in TIPS (Treasury Inflation-Protected Securities) laddered to match her spending year by year, and $200,000 in equities. Her consumption through age 75 is hedged — the TIPS deliver inflation-adjusted income regardless of what markets do. Her equity sleeve provides upside for late-life spending and healthcare contingency. Her funded ratio is 104% on the hedged portion. The 10% "failure" probability in the Monte Carlo comes entirely from extreme longevity scenarios past age 90 — and she knows that because the liability framework tells her exactly where the residual risk sits.

Investor B holds $1,000,000 in a 60/40 portfolio of US equities and nominal bonds. Nothing is hedged. Her spending is fully exposed to inflation, real-rate shifts, and equity drawdowns. If inflation runs at 4% for a decade, her purchasing power erodes by a third. If real rates fall 200 basis points, her liability grows by 25% while her nominal bonds barely move. Her funded ratio is 73% on a liability-matched basis — and falling. The same "90% probability" from the Monte Carlo conceals a fundamentally different risk position.

Monte Carlo gives both investors the same number. The funded ratio shows they are in completely different situations. One has hedged 80% of her consumption. The other has hedged nothing. The simulation cannot see the difference. The balance sheet can.

What You Lose When You Use Probability Instead of Measurement

The cost of Monte Carlo is not that it is wrong. It is that it is uninformative. A probability of success tells you one thing: in a set of simulated scenarios, your portfolio survived some percentage of the time. It tells you nothing about:

  1. Which risks are hedged and which are not. The funded ratio decomposes your portfolio into hedged and unhedged components. Monte Carlo gives you one number.

  2. What kind of loss you are exposed to. A funded ratio framework distinguishes between self-correcting losses (good beta — the market falls, but future returns rise) and permanent losses (bad beta — the money is gone). Monte Carlo treats all losses the same.

  3. How inflation changes your picture. Your spending is real. If your portfolio is nominal, inflation erodes your funded ratio every year — silently, without the Monte Carlo noticing, because the simulation denominated everything in nominal dollars.

  4. What to do about it. "92% probability of success" gives you no direction. An 82% funded ratio with a 15-year duration gap and a negative inflation beta gives you a precise diagnosis: extend duration, add inflation-linked assets, close the gap.

The Monte Carlo is a thermometer that reads in the wrong units. Your temperature is in Celsius and the thermometer reads Fahrenheit — except there is no conversion table, because the simulation's assumptions are not yours.

The Question Worth Asking

I said in 1969 that the objective of financial planning is to maximise the utility of lifetime consumption, not to maximise terminal wealth. Fifty-seven years later, the retail finance industry still optimises for terminal wealth. "Will my portfolio survive 30 years?" is a terminal-wealth question. It asks whether the pot reaches zero. It does not ask whether the person's life is funded.

The right question is: "How many years of the life I want are paid for?"

That question requires a balance sheet, not a simulation. It requires measuring your spending as a liability — with duration, inflation sensitivity, and mortality weighting. It requires comparing your assets to that liability on every risk dimension that matters. And it produces a number — your funded ratio — that tells you where you stand, what is hedged, what is exposed, and what to do next.

Pension funds have done this for fifty years. It is time you did it too.

A funded ratio is a measurement. A Monte Carlo probability is a performance. One tells you where you stand. The other tells you what happened in 10,000 imaginary histories.

Next week: "You Have a Six-Figure Asset You're Not Counting" — what Social Security is actually worth in present value, and why leaving it off your balance sheet is the most expensive mistake in personal finance. Subscribe so you don't miss it.

This is Post 2 of The Funded Ratio — a series on amifunded.com applying pension fund mathematics to individual retirement portfolios. Built on the Campbell-Viceira (2002) intertemporal asset allocation framework and the lifecycle consumption theory (Merton, 1969; 1971) used by pension funds, endowments, and sovereign wealth funds worldwide. No financial advice is given. All methods use publicly available academic research and market data.