Limitations — What It Does *Not* Do

Understanding the edges of the model is as important as understanding its features. RetireRange is a model, not a crystal ball — every result is a projection built from the assumptions you provide, and modeling is a judgment call, not a prediction. Small changes in your inputs can produce meaningfully different outcomes. We’d rather be upfront about the limits than imply a precision the math can’t deliver.

Markets and returns

• "Normal" randomness understates extremes. In Stochastic (Parametric) mode, returns are drawn from a bell curve. Real markets have "fat tails" — crashes and booms happen more often than that curve predicts. New scenarios default to block bootstrap, which resamples actual market history and avoids this; the Historical backtest is a further cross-check. (Bootstrap has its own limit: it can only replay patterns that 1928–2025 actually contained.)
• Your asset mix doesn’t change by itself. Your stock/bond/cash allocation comes from how you build your accounts, and RetireRange doesn’t shift you toward bonds as you age unless you ask it to, with rebalancing and glide-path waves (below).
• Rebalancing is off unless you turn it on, and the target never changes. By default nothing moves value between your accounts, so each one simply grows at its own return and the mix drifts — a plan that starts 60/40 typically ends far more equity-heavy. The Rebalancing card on the Strategy tab steers it back to a target once a year, and it only moves dollars between accounts that share an owner and a tax type. A household with all its stocks in Traditional and all its bonds in Taxable has nothing it can rebalance. You can change the target later in the plan with glide-path waves, but each wave is a step (no gradual slide between mixes), the schedule follows your household’s oldest person’s age (so with a wide age gap the younger spouse’s accounts de-risk on the older one’s timeline), and waves are keyed to age only, never to market conditions or portfolio value.
• Cash-reserve recovery refill is simplified. Recovery refill (a Strategy tab option) rebuilds your cash cushion after a market rebound by selling some stocks back into cash. Three simplifications to know:
  • It only refills within the same account type. Cash is topped up by selling stocks held in the same tax bucket — Traditional, Roth, or Taxable — never by moving money between buckets, since that would be a real withdrawal or rollover with tax consequences the model doesn’t handle. So a cash account with no stocks in its own bucket (a Roth holding only cash, say) can’t be refilled — its shortfall just waits.
  • It refills all at once, not gradually. In the month a recovery is detected, it tops the cushion back to full in a single step — there’s no easing back in over several months.
  • The stock sale is tax-free in the model. Selling stocks to cash inside a Taxable account for a refill books no capital-gains tax (see the capital-gains note under Taxes below). Inside Traditional/Roth accounts the shift is genuinely tax-free anyway. This is different from the annual Rebalancing card, which does realize and tax gains when it sells inside a Taxable account.
• Correlation is held constant. Stocks and bonds use a fixed relationship; in real crises, assets that normally move independently can fall together.
• Inflation is a single index (with an optional separate medical rate). It doesn’t model a sustained regime shift like the 1970s — the historical tools capture that better.

Taxes

RetireRange estimates taxes for planning direction, not filing precision. Most simplifications lean conservative (they tend to overstate your tax):

• Taxable-account withdrawals are treated as ordinary income — real long-term capital gains get lower rates, so this overstates that tax.
• Roth withdrawals are treated as income-tax-free, and the 5-year rule isn’t enforced: any withdrawal at 59½ or older counts as qualified, even from a Roth opened recently. That can understate tax, the exception to this list’s conservative lean. A pre-retirement withdrawal from a Roth before 59½ is taxed as income and penalized, unless you tick Roth basis on a Roth IRA. But Roth draws anywhere else in the plan aren’t taxed or penalized at all, even before 59½.
• Inherited-account taxes are a flat reduction, set per account. Each account’s Beneficiary setting (spouse or non-spouse) drives how any leftover balance is taxed to heirs in your after-tax legacy: a spouse inherits Traditional and HSA balances with tax-advantaged status preserved (no reduction), while a non-spouse heir takes them as ordinary income — Traditional under the SECURE 2.0 10-year rule, an HSA as immediate Pub 969 income. Roth passes tax-free and Taxable steps up either way. The simplification is that this is applied as a single multiplier on the ending balance, not a year-by-year simulation of the heir’s own drawdown — so it captures the direction of the Roth-vs-Traditional-vs-HSA difference, not the exact timing of the heir’s tax. Because it taxes the whole inherited balance at once — rather than letting the heir spread it across the SECURE 10-year window into lower brackets — it can overstate the heir’s tax (a ceiling, not a precise bill).
• Not modeled: capital-gains rates, qualified dividends, the Alternative Minimum Tax, the Net Investment Income Tax, Qualified Charitable Distributions, tax-loss harvesting, asset-location optimization, and estate tax.
• State tax covers all 50 states + DC, with each state’s Social Security tests and its retirement-income and military-pay rules applied in every state (see Taxes). The simplifications: a handful of narrower state breaks aren’t applied (Utah’s $450 retirement credit and Montana’s 65-and-over subtraction, for example); age tests use the youngest person in the household; exclusion amounts and income limits don’t rise with inflation; and county/local taxes aren’t included. All of those overstate tax. Two cut the other way: New York’s benefit recapture for higher incomes isn’t modeled and New York City’s income tax isn’t included, so a high-income New York plan understates state tax.
• We model current law, indexed for inflation — not future tax policy. Brackets, rates, and the standard deduction start from today’s law and inflate forward each year (as the IRS does). The engine does not predict future statutory changes — a rate change, a scheduled sunset, a new law — so, unlike the simplifications above, this one isn’t necessarily conservative; it simply assumes today’s rules continue. We keep the current-law tables current as they actually change.

Social Security

• A hand-entered SSA estimate assumes you keep working until you claim. Retire earlier and your real benefit may be lower, because the missing years drop out of your earnings record. If you’ll stop before you claim, either trim the estimates you enter — they’re your inputs — or import your SSA statement XML and use Recompute PIA from earnings history, which rebuilds the benefit from your actual earnings record under a stop-working age you choose. See Social Security. The recompute doesn’t model the Windfall Elimination Provision or Government Pension Offset, the disability freeze on the indexing period, or death before 60.
• The survivor benefit uses the SSA age-graded factor — 71.5% at age 60 rising to 100% at the survivor’s full retirement age (full 100% at or after FRA) — and the survivor keeps the larger of the two benefits. Finer survivor-rule edge cases aren’t separately modeled.
• Working while collecting is modeled, with a few simplifications. RetireRange applies payroll tax to earned income, runs the Social Security earnings test (including the benefit increase at full retirement age that gives most of the withheld money back), and can credit your earnings toward a higher future benefit. The edges, most of which lean conservative:
  • Only the annual earnings test is modeled. SSA also offers a special monthly test in your first year of retirement, so if you stop working mid-year, the model may withhold more in that first year than SSA would.
  • A spousal benefit is restored at the worker’s ratio. At full retirement age, any spousal step-up is raised by the same proportion as the worker’s own benefit, although spousal benefits follow their own reduction schedule. It’s an approximation.
  • Future-benefit credit requires an SSA import. Earned income raises your benefit only through Recompute PIA from earnings history on the Household tab. If you enter your benefit estimates by hand, a modeled part-time job doesn’t change them.
  • Inflation-tracking pay is credited flat. An earned stream whose growth rate is left blank counts toward your benefit in today’s dollars, without real wage growth, so the credit is mildly understated.
  • Household-scoped earned income is neither tested nor credited. Both rules belong to a person, so earned income assigned to Household skips the earnings test and adds nothing to anyone’s benefit. RetireRange shows an amber note on the Accounts tab when this happens; name the earner to fix it.
• RetireRange doesn’t predict policy changes — it projects your stated benefit forward. It doesn’t forecast reforms that haven’t been enacted. It does, though, keep the model’s current rules up to date: we maintain the tax brackets, IRMAA thresholds, RMD tables, and Social Security parameters as the law changes. See Social Security.

Costs and other items

• No investment fees, fund expense ratios, or trading costs are deducted. For low-cost index funds this is immaterial; for higher-fee or actively managed accounts, lower your return assumptions to compensate.
• Not modeled: long-term-care insurance; reverse mortgages and HELOCs; and buying a new annuity mid-plan (an existing annuity income stream is modeled). Cash-value life insurance is partial-support — the death benefit and tax-free policy-loan income are modeled, but lapse, MEC, and surrender charges are not. Real estate is modeled as an appreciating asset (with optional rental income and a sale event), not merely a rental-income source.
• A sold asset’s full proceeds become investable. When you model selling a large asset — a home, a rental — the entire net amount you enter is deposited into your portfolio and becomes available to fund retirement spending. If you’ll actually spend part of it (buying a replacement home, a big purchase, a gift), enter only the cash that will truly reach your portfolio, not the gross sale price. See downsizing your home.
• Asset classes are limited to Stocks, Bonds, Cash, and VEBA — international, small-cap, REITs, and alternatives aren’t separate categories. If you hold assets with a materially different risk/return profile and think it moves your outcome, approximate it by adjusting your stock (or bond) return assumptions — but carefully: you’re standing in for a whole asset class with a single number, and it’s easy to flatter or punish the plan by overdoing it.
• RMDs pool per person and tax type, not per plan. Required distributions are aggregated across everything you hold of the same tax type — matching the IRS rule that lets any Traditional IRA satisfy the whole Traditional IRA requirement — and they’re drawn through the same machinery ordinary spending uses, so your cash reserve and downturn-aware ordering apply to them. The one looseness left: two separate employer plans of the same type held by the same person also pool, where the IRS would require each to distribute its own. The dollars distributed and taxed are identical either way; only which sleeve gets sold differs. Surplus RMD dollars above your spending are redeposited into a Taxable account. See Where the cash reserve actually lives.
• No still-working exception for RMDs. If you’re still employed past your RMD age and own 5% or less of the company, the IRS lets you delay RMDs from that employer’s 401(k) until you retire. RetireRange doesn’t apply this exception: RMDs start at the standard age on every Traditional account, whether or not you’re still working. If this applies to you, your plan shows taxable withdrawals starting earlier than they would. That leans conservative, but it can overstate taxes and IRMAA in those working years. See RMDs.
• Optimizer precision: every candidate is now scored against identical simulated markets — one seed for the whole run, so the gap between two candidates reflects the strategy rather than two different rolls of the dice. Sampling error still applies to the absolute numbers, as it does everywhere, so a very close ranking is still a close ranking. And one gap remains: with stochastic mortality on, lifespans are drawn from the same random stream, which lets two candidates end up with slightly different death timing at the same simulation. Switch mortality to deterministic when a ranking is close, and confirm your finalists on the Compare tab.

The bottom line

Results are based on modeling and assumptions and are for educational and entertainment purposes only. Past performance does not guarantee future results. Consult a qualified financial planner for advice specific to your situation.

RetireRange is built to make you a sharper, better-informed participant in your own retirement planning — a powerful way to explore the decisions that matter and to prepare for the conversation with your advisor. What it doesn’t do is give advice. It helps you see how particular decisions and strategies are likely to affect your future financial position, but it doesn’t carry the training, credentials, or hard-won experience a financial professional brings to your specific situation — the judgment about which trade-offs are right for you. Treat every result as one well-reasoned input to your thinking, never the final word.

For how to read the numbers themselves, see Understanding your results.