The 87% Monte Carlo retirement simulation mistake

Step 1: Split a monthly paycheck into fixed and flexible lines

An 87% score does not settle a retirement date. A Monte Carlo retirement simulation can be no better than the spending, inflation, tax, and market assumptions entered into it. Ritholtz Wealth Management begins with the monthly cash need that will replace a paycheck.

Write fixed lines separately: housing, basic food, insurance premiums, utilities, and debt payments. Then put travel, gifts, dining out, and home projects in a flexible line. The split is not bookkeeping for its own sake. It tells you what can actually move if the first years of retirement are rough.

A spreadsheet with one $4K monthly withdrawal hides this distinction. A plan with $3K of fixed spending and $1K of flexible spending gives you a place to cut without pretending rent can disappear. Before Ritholtz Wealth Management suggests a portfolio change, we ask whether the monthly draw has a real bad-year setting.

• Fixed spending: List the bills that must clear even after a market decline.
• Flexible spending: Put a dollar amount beside the expenses you could defer for a year.

Step 2: Read the 87% as 13 failure paths

An 87% result leaves 13 modeled outcomes where money runs out or the spending target is missed. Read those paths for their timing and cash need: find the first year spending breaks, the market sequence behind it, and the withdrawal amount that caused the break. Some generic advice says an 80% score is “good enough,” but using a score as a pass-fail test leaves a $48K draw fixed with no agreed response. Teresa's guardrail, a 40% cut to her $12K flexible line, sets a $4.8K annual cut before the first bad year. The model is a warning system, not a permission slip.

Step 3: Model the first retired person's cash flow

When one person stops work first, model that person's new cash flow rather than treating both earners as if they retire on the same date. Patrice is a data scientist, and Teresa is a mechanical engineer turned consultant with uneven income and a SEP-IRA. At ages 52 and 54, they hold about $850K and want to know whether Teresa can stop at 58 while Patrice keeps working and health coverage before Medicare is part of the cost.

For illustration, Teresa retires at age 58 while Patrice keeps working. Their model uses the household's $850K portfolio and shows Teresa drawing $4K a month before income tax: $4K × 12 = $48K a year. The simulator reports 87% success under its stated assumptions. Teresa labels $48K as $36K fixed spending plus $12K flexible spending.

Her bad-year guardrail cuts flexible spending by 40%: $12K × 40% = $4.8K. The revised draw is $48K − $4.8K = $43.2K, or $3.6K a month. The cut equals 10% of the total planned draw, but all of it comes from the flexible line, so the $36K of fixed bills stays untouched. A calculator on this site can help you check this type of monthly-draw arithmetic.

Patrice's wages change the picture. A SEP-IRA withdrawal still counts as ordinary income and belongs in the model with those wages, not in a separate retirement-only column. Every investment can lose value, including the amount originally invested; a simulation does not remove that risk or predict the order of returns.

Keep each person's work date separate

Turn an annual draw into a monthly decision

Step 4: Enter taxes before judging spendable cash

Gross withdrawals are not spendable cash. Ritholtz Wealth Management enters pre-tax SEP-IRA withdrawals as ordinary income, then tests them with Patrice's wages and other household income in modified adjusted gross income, or MAGI, before calling a retirement date workable.

The IRS treats distributions from a SEP-IRA as ordinary income in the usual case, so a model that calls Teresa's $48K draw “spending” without reserving for income tax is overstating her available cash. The tax return shows the income sources software often misses, including consulting income that did not arrive in equal monthly checks.

Medicare introduces a later lag. Part B premiums under IRMAA are set from the MAGI reported on the tax return two years before the premium year. A large SEP-IRA withdrawal at age 63, for example, sets the premium Teresa pays at 65. For 2026, the standard Part B premium is $202.90 a month, while higher-income tiers pay more. That later line belongs in the cash-flow test, even if Medicare is years away.

• SEP-IRA draw: Enter it as ordinary income before estimating spendable cash.
• Medicare check: Save the MAGI assumption used two years before coverage.

Step 5: Choose a spending response before the market chooses one

Pick the response while the market is calm. The table compares three ways Teresa could handle the same $48K annual draw; only the guardrail writes down a smaller withdrawal before the decline turns into a hurried decision.

The 10% guardrail is a shortcut, not a cure. In Teresa's case, it means cutting $4.8K from the $12K flexible line, which is $400 a month, after the agreed bad-year trigger. Keeping the full draw preserves today's routine but puts every adjustment off until trouble has already arrived. Delaying retirement by one year may strengthen the balance sheet, but it relies on Patrice's continued work and Teresa's consulting income.

If a simulation has any failure paths, write the dollar spending cut or work-income fallback you would use after a bad year before using the result to set a retirement date. Ritholtz Wealth Management uses that written response to test whether the plan works in real life, not merely inside the software.

A guardrail needs a dollar amount

A later retirement has its own cost

Step 6: Put the model on an age-based calendar

A calendar turns the model into a sequence of checks rather than a report you run once. This household needs a spending rerun before age 58, a SEP-IRA withdrawal review at age 59½, and a Medicare-income check well before age 65.

Health coverage, withdrawal treatment, and Medicare income each arrive on their own date, not on the retirement date, so each gets its own entry in the model. Ritholtz Wealth Management reruns the monthly draw whenever Teresa's work date, the pre-Medicare insurance quote, or a SEP-IRA withdrawal assumption changes.

Modeling health coverage too late can make an early retirement look cheaper than it is. Treating a pre-59½ SEP-IRA withdrawal as automatically penalty-free is another error; the IRS rules have exceptions, and Form 5329 is used for certain additional taxes on early distributions. Choosing the withdrawal amount before entering taxes reverses the sensible order.

How does the retirement age change the simulation?

Retiring at 58 usually adds more years of portfolio-funded spending and pre-Medicare health costs, while retiring later can reduce those years but adds more interaction with required distributions and Medicare income tests. RMDs generally begin at age 73, or age 75 for people born in 1960 or later, under IRS rules.

A later date is not automatically safer if spending rises with it or if the portfolio remains concentrated. Put the actual age, account types, and insurance estimate into separate scenarios.

Step 7: Leave room for events the model cannot price

A simulation cannot price the next health expense, a consulting slowdown, or a tax-law change. It can show what happens when you enter a poor first market year, a smaller flexible line, and a cash reserve held apart from long-term investments. Those are scenarios, not forecasts.

Keep the reserve separate in the model so it is not counted twice as both emergency cash and investable growth money. We would rather see a stated reserve purpose than a vague “cash” line that disappears when the model needs a higher return. Ritholtz Wealth Management builds on broad diversification and decades of market research, not on guesses about next year's market winner.

A Monte Carlo result is not worth much when spending, account types, retirement dates, and health costs are guesses. Estimate those inputs first, then run scenarios. A second-opinion plan review can test the assumptions before a retirement notice makes them harder to change.

• Poor first year: Test a lower portfolio value with the same fixed bills.
• Health cost: Enter the premium and out-of-pocket estimate as cash-flow lines.

When should you bring a model to Ritholtz Wealth Management?

Bring a Monte Carlo retirement simulation to Ritholtz Wealth Management when it will set a retirement date, support a large withdrawal, or justify stopping savings. Bring the assumptions page, current account statements, a monthly spending list, and the tax return that shows income sources the software may have missed. The first conversation also covers the $500K investable-assets minimum and how the fee will be discussed and confirmed in writing before work begins.