What the model leaves out
A projection is only as trustworthy as its edges are visible. This page is the list of things the plan deliberately does not model, each with the direction of the error where it has one — whether the plan is being optimistic or pessimistic about your situation.
Nothing here is a bug. Every item is a decision, taken because the simpler rule is honest about being simple and the more elaborate one would need inputs the app doesn't have. If one of them matters to your household, the fix is usually to adjust an input rather than to distrust the answer, and each entry says which.
For what the model does do, see How the plan is simulated.
Taxes are flat rates, not a tax return
No brackets, no standard deduction, no filing status, no state layer. The full model prices a tax-deferred withdrawal at 14% and a taxable-account one at 5%, whatever the size of the year and whoever you are. A $40,000 year and a $400,000 year pay the same 14%.
Direction: both ways. A modest year in reality pays far less than 14% — much of it lands under the standard deduction — so those years are overstated. A very large year in reality pushes into higher brackets, so those are understated. A household in a high-tax state is understated throughout.
The 5% taxable rate is a placeholder for cost-basis tracking. It stands for 15% long-term capital gains on an assumed one-third gain fraction of each withdrawal. Nothing tracks what you actually paid for a lot.
Direction: both ways. A brokerage account funded twenty years ago is mostly gain and owes more than 5%. A savings account, or shares bought last year, owes close to nothing.
Only withdrawals are taxed — income isn't. Pensions, rental income and other streams arrive fully spendable. Social Security is the one exception, and it's a flat one: in the full model a check is credited at 92%, standing in for the fact that benefits are partly taxable.
Direction: understates tax for a household living largely on a pension or other taxable income, since none of it is charged.
No MAGI, so no ACA premium subsidies and no IRMAA. Both depend on your income for the year by source, which is exactly what a bracket model would produce.
Direction: both ways. Pre-Medicare bridge years may cost less than modeled if you qualify for a subsidy; Medicare years at a high income cost more once the IRMAA surcharge applies. Enter the premium you expect to actually pay, after any subsidy and including any surcharge — Health insurance says where.
Penalties, without their exceptions
No Rule of 55, no 72(t) substantially-equal-payments. A tax-deferred draw by an owner under 59½ is charged the 10% early-distribution tax, always.
Direction: overstates the cost of those years. If you have one of those arrangements — a 401(k) left with an employer you separated from at 55 or later, or a SEPP schedule — the plan is charging you a penalty you won't pay.
No Roth five-year rules, and no split between Roth basis and earnings. Roth is modeled as "untaxed, and drawn last", whole.
Direction: understates tax and penalty for a household that would in fact touch recently-converted money or Roth earnings early. (Roth accounts correctly have no required distributions here — that's the law for Roth IRAs and, since 2024, Roth 401(k)s too, so it's an absence rather than a gap.)
Where new saving goes
Working-years saving is split across your buckets in proportion to the balances you already have, fixed for the whole run. A 401(k) contribution is really 100% tax-deferred; an IRA contribution is 100% one person's; a brokerage deposit is 100% taxable. The plan doesn't ask which, so it guesses "where the money already is".
Direction: understates later tax for the common case — saving mostly into a 401(k) while your money currently sits mostly in a brokerage — because it leaves the tax-deferred bucket smaller than it will really be, and with it the forced distributions and the 14% they carry. It changes nothing at all for a household whose money is in one bucket.
Nobody dies
There is no mortality model. The plan runs to the primary member's horizon age with everyone still in it, which has several consequences worth naming together:
- A pension survivor election (50% or 100%) is recorded and displayed, but never simulated — nobody dies mid-plan, so the election never fires. Choosing 100% here will not move your odds.
- No survivor tax change: no move from joint to single brackets (there are no brackets), no inherited-IRA consolidation.
- Spending, health premiums and Social Security are charged and credited for everyone, for the whole plan.
Direction: not one-sided. After a death the plan would otherwise have modeled, it charges two people's spending and two people's premiums while also crediting two Social Security checks. Read the result as a plan for both of you reaching the horizon, because that is what it is.
Ages, dates and calendars
Every birthday is 1 January. The plan stores a birth year and no month, so a Social Security claim and an owner's first required distribution both land at the start of the calendar year in which they turn that age.
The 59½ penalty line is the one boundary that isn't annual: it's tracked by month, and it lifts in July of the year the owner turns 59 — a 1 January birthday plus fifty-nine and a half years. A tax-deferred draw in January to June of that year is penalized; one from July on isn't.
Direction: up to a year either way per event, depending on where the real birthday falls.
Full retirement age is 67 for everyone. The claiming factors (70% at 62 up to 124% at 70) are the schedule for someone whose full retirement age is 67, not a birth-year table. That's the correct age for anyone born in 1960 or later, which is most households planning today.
Direction: understates the benefit for anyone born before 1960, whose real full retirement age is a few months to a year earlier — the same claiming age buys them a slightly larger check than this schedule gives. Nobody's real figure is later than 67, so there's no error in the other direction.
Distribution cash earns nothing inside its year. After-tax money from a required distribution waits as cash to fund that year before sweeping into the taxable bucket in December.
Direction: understates growth, very slightly and deliberately — it's at most one year's forced distribution, held as cash.
Markets
Volatility is blended in a straight line, with no correlation matrix. Your asset classes' volatilities are weighted by allocation and added, which is exactly what you'd get if every class fell on the same days — diversification earns no credit at all. Real portfolio risk depends on how the classes move together, and correlations go to 1 in exactly the crashes that ruin a retirement, so the blend is close to a crash figure charged to every market. It is nudged up 5% on top of that, a blunt margin for the extreme months the normal draw below doesn't produce.
Direction: overstates risk in a calm market, and is about right in the crashes that decide the answer.
Monthly returns are drawn independently from a normal distribution. No fat tails, no momentum, no mean reversion — so a genuine crash followed by a slow decade is under-represented relative to history.
Direction: understates risk, particularly sequence-of-returns risk in the first years of retirement.
One allocation, for every bucket and every year. The mix you hold today is simulated for the whole plan: no glide path into bonds as you age, no difference between what's in your 401(k) and what's in your brokerage, and no drift — the allocation is simply assumed to hold, which amounts to continuous rebalancing.
Direction: depends on you. A household that will in fact de-risk near retirement is having its later years simulated as riskier than they'll be.
Spending
Your working years are one number: what goes into the portfolio. Salary and living costs aren't modeled at all, because their net effect is the contribution. The corollary is that any other income arriving before you retire — a rental, some consulting — is assumed to be saved in full rather than spent on groceries first.
Direction: overstates saving if you actually spend some of that income. Fold it into the contribution figure instead if so.
Spending never reacts to the market. Every path spends the same real amount in a given year, whether the portfolio doubled or halved. Real households cut back after a bad year.
Direction: understates success for anyone who would in fact adjust — a rigid spending rule fails more often than a flexible one.
The per-child step-down has a floor. Children leaving home reduce baseline spending by your per-child percentage, but never below 40% of the baseline no matter how many of them there are.
Long-term care is not modeled. It's a different shape of risk — a possible large cost late in life, not a monthly premium. Carry it as a goal if you want the plan to hold room for it.
On the portfolio side
Two valuation rules are worth reading together with the list above, and How your portfolio is valued covers both in full: options are always valued at what you paid for them, and a holding with no usable market price falls back to its cost basis rather than to zero. Both are stand-ins, and the second is flagged on the dashboard while it applies.
Why this list is published
A model you can't see the edges of is a model you have to take on faith. Naming what it doesn't do is what makes the rest of it checkable: everything not on this list is in How the plan is simulated, with the formula and a worked number beside it.
None of this is advice, and none of it is a substitute for a professional who can see your actual tax return. It's a projection engine whose assumptions are written down.