What-If scenarios are pre-built stress tests. Each one perturbs your saved simulation and re-runs the engine. They are non-destructive overlays. Your baseline inputs don't change.

Open them from What If? in the sidebar. That takes you to the What-If Scenarios page: pick a scenario from the list on the left, set it in the panel on the right, then Save and run to re-run the Proof with it applied.

The single-knob scenarios

Equity Drop at Retirement

A one-time equity-only haircut applied in your retirement year. Use it to model a sequence-of- returns shock right when you start drawing down. Slider range is 0% to -60%.

Asset Return Lower

Persistently reduces equity returns across the entire simulation. This is the "what if the long-run average is just lower than history?" knob. The entire return distribution shifts down. Slider range is 0% to -10%.

Inflation Higher

Adds (or subtracts) a constant percentage to inflation for every year of the simulation. Slider range is -10% to +10%. Only one inflation scenario can be active at a time.

Inflation Spike

Adds an inflation bump to the first five years only, then returns to historical inflation. Models a near-term shock without permanent damage. Slider range is -10% to +10%.

Social Security Reduced

Cuts every Social Security payment by a percentage. Models the often-discussed scenario in which trust-fund depletion forces a benefit reduction. Disabled if you don't have a Social Security adjustment configured. Slider range is 0% to -75%.

Capital Gains Increase

Replaces the long-term capital gains tax rate (currently 15%) with a different value. Useful for stress-testing portfolios that depend heavily on a taxable brokerage drawdown. Slider range is 0% to 50%.

Sequence-of-returns stress test

The one scenario that takes more than a single number. It replaces the first few years of returns with flat rates and then resumes your normal returns, which is the standard way to ask "what if the bad years land first?" The first decade of retirement drives portfolio survival far more than the average return does, so the same long-run average delivered in a worse order can break a plan that otherwise works.

• Years to override (0 to 10) is the switch. At 0 the scenario is off and the rate inputs are disabled, whatever they say.
• Equity, Bond, Gold and Inflation rates are each independently on or off. A rate switched off leaves that asset class on your Parameters tab settings for the whole run. Equity and Bond start on; Gold and Inflation start off. Each has a Use historical average button that fills in the long-run figure: 9% for equities and 4.5% for bonds, the nominal compound annual growth of the same series the simulation replays since 1871 (Shiller's S&P composite with dividends, and the 10-year Treasury), 5% for gold (1928 onward) and 3% for inflation (CPI since 1926). These are nominal figures, before inflation, because inflation is its own rate here; the familiar "7% for stocks" is the real return with inflation already taken out.
• Each rate is a total return, not a price change. The overridden years pay no separate dividends, so a taxable brokerage funds spending by selling during the window rather than from dividend cash.
• Overriding inflation here conflicts with the two inflation scenarios above, so you can use one or the other, not both.

Career gap

"Can I take a couple of years off work, or survive a spell without it?" Tick one or more income streams, say when they stop and for how long, and the plan runs as though those jobs simply paid nothing for those years. A sabbatical, a layoff, and a couple travelling together are the same mechanic with different inputs, so they share one row. The spending change applies once to the household, however many jobs pause.

• Everything that rides on the job pauses with it. Payroll deferrals, the employer match and FICA all stop for the gap and resume when the job does. Your Cash Flow Priorities and any Solo 401(k) link stay attached — the job is switched off, not deleted, so nothing has to be set up again afterward.
• The job has to be running during the gap. If the job you pick ends before the gap starts, or starts after it ends, the row shows an error and neither Save nor Compare to baseline will run until you pick another job or move the gap.
• Returns to work afterward is on by default. Turn it off and the job ends for good the year before the gap starts, which is the layoff-with-no-comeback case.
• Spending change during the gap is one optional slider covering both belt-tightening and a sabbatical that costs more. It is a percentage of what your working-years period spends and lowers or raises Base Spending in the gap years only; the Proof's Cash Flow card annotates Base Spending with the change. It needs a working-years period on your Base Living Expenses, and the slider is disabled until you add one.
• The gap has to start before your retirement year — a job that has already stopped cannot pause.

Downsizing your home

"What happens if I downsize?" Pick your primary residence, say when you sell, and choose what you do next. The sale is modelled exactly as a manual one would be — net proceeds after selling costs and any remaining mortgage, with the primary-residence capital-gains exclusion applied — and whatever is left after the move lands in your portfolio.

• Buy for cash or buy with a mortgage size the new home as a percentage of the sale price, not as a dollar figure. That is deliberate: nobody can say what a house will cost in 2040, but "about 60% of whatever mine fetches" holds in every scenario the engine runs. The carrying costs follow the new price — property tax and maintenance are rates, and insurance scales with the smaller home.
• Rent adds an annual rent from the sale year onward, rising with inflation, and no property costs at all.
• The figures under the editor are quoted at today's value. Each run sells at whatever the property is worth in that year, so the dollars move but the shape of the trade does not.
• Your own Real Estate adjustment is left alone. If you decide to go ahead, set the sale year on the property itself.

Can I retire earlier?

The last row on the list is not a scenario. It is a solver: it tries every candidate retirement year from next year through the one in your plan, runs your plan at each one, and lists every year that scores at least 80%, each with the success rate your plan actually reaches there. Two years can sit a few points apart, so the list shows the real spread rather than one line per round target. Below the list, a slider asks for any success rate you like and names the earliest year that meets it, read straight from the same sweep, so moving it costs nothing. If your plan misses a target even at its current year, the sweep keeps going, one year at a time, until the youngest person turns 75 or the plan runs out of years, so "working until 2039 reaches 90%" is an answer it can give.

• Every year is measured, not guessed. The curve of success rate against retirement year is not smooth: a health-insurance subsidy cliff or the first year of required distributions can make an extra year of work score worse. Because the solver measures every year, it cannot miss a better year hiding behind a worse one, and it tells you when working longer scored worse.
• What moves with the year. A Job, Take-Home Salary or Side Hustle that ends the year before retirement (or in the retirement year) ends the year before the candidate instead, and payroll deferrals and the employer match follow it. Anything you pinned to Retirement Year when you set it up, such as an ACA Healthcare start or a home sale, moves with it too. Everything else stays where you put it: Social Security claiming, pensions, and any adjustment dated to a calendar year. A job that already ended earlier, or that runs well past retirement, is left alone.
• Your spending time periods move too. If your Base Living Expenses are split into time periods, the retirement periods slide to the candidate year keeping their lengths, and the working-years period stretches or shrinks to meet them. A boundary you drew before your retirement year stays put, on the same reasoning as Social Security and pensions: you pinned it to a calendar year deliberately. If retiring early would leave a period with no years in it, that period is dropped and the list of changes says so.
• The plan length does not move. Your plan runs a fixed number of years from today, so a later retirement means fewer retirement years inside it, exactly as if you changed the year yourself on the Inputs tab.
• Your saved What-Ifs are included. The solver runs the plan as it stands. If you have saved "asset return lower", it finds the retirement year under that stress.
• Apply sets the retirement year on your plan and makes the same job and date edits the solver made, after showing you the list. It then re-runs the Proof. In Explorer mode you can solve but not apply.
• Some plans cannot be solved. A Risk-Based Guardrails spending plan and Monte Carlo returns each make a single evaluation too slow to repeat for every year, and a plan with no retirement spending scores the same in every year. The row explains which applies. Solves run in the background and share one worker slot with Monte Carlo runs, so starting one cancels a Monte Carlo run in progress. Results are kept until the plan changes; opening the row later shows the last answer without solving again.

What if I flex my spending?

Puts a floor and a ceiling around your retirement base spending, so a good decade lets you spend a little more and a bad one asks a little less. Pick a band of 10% or 20% either way, or drag the two handles on the band slider; the mark at its center is your base spending in today's dollars and moves with the change slider, and each handle shows the dollar figure it sets. On an Inflation-Adjusted plan the band turns it into a Variable Spending plan for the run; on a plan that already has a floor and ceiling, such as Variable Percentage Withdrawal or Percent of Portfolio, the band tightens or widens the ones you set. The same row also lets you try a permanent change to the base itself, from 50% less to 30% more; on a Percent of Portfolio plan the change scales the withdrawal rate instead, and on VPW there is no base to change. The change is applied first and the band around the result. Floors and ceilings are in today's dollars and rise with inflation, like the base.

What if I used guardrails?

Swaps your spending plan for Guyton-Klinger guardrails for the run: spending is cut by the percentage you set when the withdrawal rate climbs 20% above where it started, and raised when it falls 20% below. Your base spending carries over. Guyton-Klinger needs a floor and a ceiling or its moves can run without limit, so the trial brings its own band, 20% either way by default and adjustable on a slider; if your plan already sets a floor and ceiling those apply instead, and a band set on the Spending flex row takes precedence over both. It needs a plan with a base spending figure, and it is not offered on a Risk-Based Guardrails plan, which already sizes spending itself. Risk-Based Guardrails as a trial is not available here yet; it runs on the background solver.

Compare to baseline

Below the editor, Compare to baseline runs your plan twice. The first run has no What-Ifs. The second run uses whatever you have set. It then shows the success rate, median ending portfolio side by side. Neither run is saved and neither changes your plan. It is there so you can see the size of a scenario's effect before you commit to it. It is available on historical-cycle plans. Monte Carlo and guardrails plans will get it later.