Your Plan's Success Rate Isn't Its Risk Profile

One number, three unrelated failure modes

Run a Monte Carlo simulation on a retirement plan and you get back a single percentage — say, 87%. It's tempting to treat that number the way you'd treat a credit score: one figure that sums up the whole picture. It doesn't. An 87% success rate means that out of however many simulated market paths were tested, 87% of them left the household with money at the end and 13% didn't. It says nothing about why the 13% failed.

That matters because the 13% that fail aren't failing for the same reason. Some of those simulated futures ran into a bad market in the first few years of retirement, while withdrawals were already being taken. Some ran fine through markets but assumed a lifespan that turned out too short for how long the household actually lived. Some kept pace for two decades and then got eaten by inflation running hotter than the plan assumed. Averaging all of that into one number is mathematically correct and, on its own, not actionable. You can't do anything about "87%." You can potentially do something about "this plan is fine unless the market drops in the first five years" — but only if you know that's the specific exposure.

This is also why two households can land on the identical success rate and be nowhere near the same actual risk. A 62-year-old with a large portfolio and a short planning horizon might be exposed almost entirely to sequence-of-returns risk. A 55-year-old coasting on a smaller savings rate with a long joint life expectancy might be exposed almost entirely to longevity risk. Same 85%. Completely different plan.

Sequence-of-returns risk: a timing problem, not a returns problem

Retirement portfolios don't just care about average return — they care about the order returns arrive in, because withdrawals are happening at the same time as the market is moving. A portfolio that drops 20% in the first year of retirement, while money is already being pulled out to cover spending, locks in losses on shares that are gone and can't participate in the recovery. The identical 20% drop in year fifteen, after a decade and a half of growth and compounding, does far less damage to the same household.

This is why "the market averages 7% a year over the long run" is close to useless for judging whether a specific retirement date is risky. Two portfolios can share the exact same average annual return over 30 years and produce wildly different outcomes purely because of when the good and bad years landed relative to the withdrawal schedule. A plan exposed to sequence risk is fragile in a narrow window — roughly the first five to ten years of retirement — and largely fine after that.

What actually helps against sequence risk

• A cash or short-bond buffer sized to cover a year or two of spending, so a down market doesn't force selling depressed shares to pay bills.
• Flexible spending in the early years — the ability to trim discretionary spending if the first few years of retirement land in a downturn, rather than a fixed withdrawal regardless of market conditions.
• Delaying retirement by even a year or two if the market has just had a strong run, since it changes which years become "year one" of the withdrawal sequence.

Longevity risk hides behind a comfortable median

Most retirement math quietly plans to a life expectancy that's really a coin-flip midpoint — by definition, roughly half the people the actuarial tables describe live longer than that number. A plan that looks completely solid out to age 90 can be genuinely under-provisioned if either spouse in a household has a real, non-remote chance of reaching 95 or 100, which for a healthy couple in their 50s or 60s today is not an edge case.

This is a distinct fragility from sequence risk, and it shows up differently: instead of being exposed in a narrow early window, a longevity-fragile plan looks fine for two or three decades and then runs into trouble late, when there are fewer options left to correct course. It's also a risk that's easy to miss precisely because the plan's median outcome — the "expected" case most people intuitively picture — looks fine. The fragility lives specifically in the tail, not the middle.

What actually helps against longevity risk

• Planning to a longer horizon than the median life expectancy — the honest version of the question isn't "how long will I probably live," it's "what happens if I live longer than that."
• Delaying Social Security, which functions as inflation-adjusted longevity insurance — the later claim pays a permanently higher benefit for as long as either spouse is alive.
• A larger cushion or spending flexibility late in retirement, since a longevity-fragile plan's problem shows up in the plan's final years, not its first.

A plan can be robust against one risk and fragile against another. "Robust" only means something once you say robust to what.

Inflation risk: the slow one

Sequence risk hits early and loudly. Longevity risk hits late and definitively. Inflation risk is the quiet one — it doesn't cause a single bad year, it compounds a small annual gap between what a household assumed costs would rise by and what they actually rise by, until the gap is large enough to matter. A category-specific version of the same risk shows up in healthcare costs specifically, which have historically outpaced general inflation and represent a real budget line that a generic inflation assumption can understate.

A plan exposed mainly to inflation risk doesn't look fragile in any single simulated year — it looks fine for a long stretch and then erodes. That's a different shape of problem than either of the other two, and it calls for a different fix: rebuilding the spending assumption with a more conservative inflation rate, or stress-testing specifically against a higher-than-historical inflation regime, rather than a cash buffer (which helps sequence risk) or a later claim age (which helps longevity risk).

A worked example: same score, different plan

Consider two households that both land on roughly the same overall stress-tested success rate when run through a simulation. On paper, "same number" reads as "same risk." It isn't.

If both households only looked at their headline success rate, they'd walk away with the same read: "solid, not perfect." But the moves that would actually strengthen each plan point in opposite directions. Household A gets real protection from a cash buffer and the flexibility to trim spending if the first few retirement years land in a downturn. That same cash buffer does almost nothing for Household B, whose real exposure is decades away and would be better addressed by delaying a Social Security claim or planning to a longer horizon. Applying Household A's fix to Household B — or vice versa — would leave the actual risk untouched while creating the comfortable illusion that something had been done about it.

What this doesn't capture

Decomposing a success rate into its component risks is a real improvement over treating it as one opaque number, but it's still a model, not a forecast. It ranks which historically-plausible risk category is doing the most damage to a specific plan — it can't tell you a market crash is coming in a specific year, that a specific tax law will change, or that a specific health event will happen. It's a projection, not a prediction. Use it to decide which lever is worth pulling, not as a guarantee that pulling it eliminates risk entirely.

How this is calculated

The breakdown above comes from a 1,000-scenario Monte Carlo simulation run against the same engine that powers the full app, using randomized historical market returns, SSA actuarial life-table data for longevity modeling, and a configurable inflation assumption — free to run, no account required. The complete math, including the specific assumptions and data sources behind each risk category, is documented in the methodology.

Common questions