How Plan Mirror calculates

Questions: support@planmirror.com

Plan Mirror turns a retirement plan into numbers you can check. A deterministic calculation engine does the arithmetic, not an AI model: the same inputs always produce the same results. The AI in the product helps you describe your plan and explains results the engine has already produced. It never adds figures of its own.

From your plan to a model

Balances, accounts, spending, income, ages and your state become explicit inputs. Anything you haven't told us is shown as an assumption you can change, never filled in silently. Spending is all-in: it covers living costs, healthcare and the income tax the engine works out for you each year.

Five ways to test a plan

No single projection can tell you whether a plan works. Plan Mirror tests the same plan five ways, and each test answers a different question.

• The steady projection. Year by year, the portfolio grows at your expected return, spending and Social Security rise with inflation, and the engine works out withdrawals and taxes. It answers: if the averages hold, what happens?
• Two stress tests. The same plan with spending 10% higher, and with returns 2 percentage points lower every year. They show which assumption the plan is most sensitive to.
• The historical replay. Runs the plan through real U.S. stock market returns from 1926, starting at every quarter that has a full plan length of history after it. Returns are adjusted for the inflation that actually happened at the time (CPI-U), so each start date gets history's real returns, including the Depression and the inflation of the 1970s. It answers: would this plan have survived the past century?
• Monte Carlo. Runs the plan through many randomly generated market paths. Your expected return is treated as the typical compound return: the median path grows at it, and volatility (15% a year by default) spreads the outcomes around it. You can also add random household expenses, such as a new roof or a car. It answers: how wide is the range of outcomes if the future averages what you expect?

Simulation results are reported as the share of paths that stay funded through your plan. That is a result under stated assumptions, not your probability of success.

Default assumptions

• Return: 6% a year before inflation. This is in line with current long-term assumptions used by financial planners. It is well below the U.S. stock market's average over the last century, which the historical replay uses.
• Inflation: 2.5% a year, applied to spending and to Social Security.
• Taxes: federal income tax on the rules for a stated tax year, and state income tax for supported states. Every scenario names the rule year it used.
• Withdrawal order: any required minimum distribution comes first, then cash, taxable accounts, traditional accounts and finally Roth accounts.

You can change the return, inflation and simulation settings for your own plan.

What the engine doesn't cover yet

• Early withdrawals. The 10% additional tax on retirement-account withdrawals before 59½ is not applied. If your plan draws on a 401(k) or IRA before then, you'll need a penalty-free route, such as a Roth conversion ladder or substantially equal periodic payments.
• Filing status. Taxes are worked out for married couples filing jointly.
• Portfolio mix. Simulations treat invested money as one stock-market portfolio. The historical replay is 100% U.S. stocks, which swings more than a typical stock-and-bond mix.
• Required minimum distributions start on an approximate schedule, not one based on your birth year.
• Everything outside the model. Real markets, tax law, health and spending will differ from any assumption.

Plan Mirror is not financial, tax, investment or legal advice. Check anything important with a qualified professional.