What's new — and why your numbers moved (August 2026)

The short version

A later round of corrections shipped in September — see [What's new (September 2026)](/help/whats-new-september-2026).

We spent early August making the math behind your forecast materially more accurate — fixing things that were wrong, and modeling real costs that were missing. Several of these change the numbers you see, in both directions. Nothing about your saved plan was altered; what changed is how honestly we project it.

Corrections — things that were wrong before

• Your investment accounts now move together. The odds-of-success simulation used to roll separate random returns for each account, which quietly made a plan split across several accounts look safer than the same money in one. Accounts now share each year's market, the way a real crash actually treats them. If your odds dropped, this is the main reason — the new number is the honest one.
• Tax brackets now grow with inflation, as they do in real life. Before, your income grew each year while the brackets stood still, inventing "bracket creep" that pushed you into ever-higher tax rates — by year 30 the tax bill could be overstated by more than 80%. This change lowers projected taxes, especially late in the plan.
• Selling investments now has a tax cost. Withdrawals from a regular brokerage account used to arrive tax-free in the projection. Now the app tracks what you paid for those investments (the cost basis, editable on each brokerage account) and taxes the gains at real capital-gains rates — including the 0% bracket many retirees genuinely qualify for.
• RMDs now use the right person's age. In a couple, required withdrawals from a 401(k)/IRA used to start when the older spouse hit 73, even for the younger spouse's accounts. Set Whose account on each retirement account under My Plan → Accounts — a younger spouse's IRA now keeps growing until their own clock starts.
• The simulation no longer assumes you die on schedule. Each run now varies how long you live around the age you entered — because the risk that matters is living longer than planned. This typically lowers the odds a little, honestly.

New realism — costs and behaviors that were missing

• Medicare IRMAA surcharges — higher incomes pay more for Medicare, in cliffs. Now included in the tax estimate and in the Roth conversion comparison.
• [Marketplace subsidies before 65](/help/aca-health-insurance) — including the income cliff at 400% of the poverty level.
• [A long-term-care stress test](/help/long-term-care) — the biggest risk most plans never model.
• [What happens after the first death](/help/survivor-modeling) — the survivor keeps the larger Social Security check, not both; taxes tighten; spending doesn't halve.
• [Spending that slows with age](/help/spending-phases) — the well-documented pattern flat projections ignore.
• Flexible spending in the simulation — the odds panel now also shows your chances if you'd trim spending ~10% after a bad market year, because nobody spends on plan through a crash. The gap between the two numbers is a decision, not a fate.

What you should do

Five minutes of review: set Whose account on retirement accounts and cost basis on brokerage accounts, and consider the new toggles under Profile & assumptions — each is off by default, so nothing turned itself on. Then re-read your run-out date knowing it's now built on sturdier arithmetic.

If a number got worse, we didn't make your retirement worse — we stopped hiding something. That's what you're paying us for. Estimates, not advice, as always.