Your plan has to answer two different questions, and it's easy to mix them up. One is what could the portfolio support? That's the return model. The other is what rule decides how much you actually spend this year? That's a choice you make, and this is where you make it.
Where to start
Which of the three to pick, and what the two numbers under the picker mean, is Which spending strategy should I pick?. This topic is the arithmetic underneath: how VPW's percentage is set, what the essential floor does, and the two places this app diverges from the published worksheet.
The three form a ladder. My expenses, as entered never reacts. My expenses, cut back in bad years nudges the flexible part when your withdrawal rate strays (the guardrail). A share of my portfolio recomputes the number from the balance outright, either rising with age (VPW, Variable Percentage Withdrawal, from the Bogleheads community) or at a fixed percentage you set, which never rises with age and so tends to under-spend early money.
The mechanism behind your choice (Guardrail (Guyton-Klinger) · Balanced, Bogleheads VPW to age 100, Fixed 4%) opens the Adjust the rule disclosure, and always reflects your own setting, not a default.
The trade nobody should hide from you
A flexible rule gives up steady spending to improve the chance the plan is funded. That isn't magic, it's arithmetic: if you always spend a fraction of what's left, there's always something left, and your spending absorbs the strain instead.
It does not follow that the money cannot run out. This app pays your essential costs even when the rule's number falls below them, and draws taxes and Medicare on top, so a balance rule can still exhaust the portfolio (see the essential floor, below). Any tool that tells you a percentage rule cannot fail is describing a different model from this one.
The home goal seekers include each candidate's recurring housing in this floor too. It raises spending only when the rule would otherwise prescribe less than the protected costs.
So a strategy comparison that only shows you success rates is worse than useless. It will always crown whichever rule cuts your spending hardest. Read both numbers together, and keep them under separate names:
- the chance all obligations are funded, beside the share of outcomes in which some expenses go unfunded, and
- what the rule chose to take off in its worst year: reduction below entered expenses.
A plan that "succeeds" at $19,000 a year is not a plan that succeeded. Reporting one number for both facts is how that gets hidden, which is why this app reports two.
Explore's comparison adds a third column, on track at an age five years into retirement: the share of outcomes that reach it still solvent and at or above the typical balance, and go on to fund everything. It is where the adaptive rules' steadier early years become visible, beside the reduction that bought them. See confidence by age.
The same split runs through the pictures. Everything the spending chart draws is a planned amount except one line, 1 in 10, as funded: what the tenth-percentile outcome actually had, after whatever the portfolio could not deliver. A planned band that holds flat while that line collapses is a rule that ran out, not a rule that held steady. In Explore's Details the same pair sits in two columns, the funded one carrying how far below the planned amount it landed.
How VPW's percentage is set
It's a loan amortization run backwards: spread the balance over the years remaining to your terminal age at the expected return of your stock/bond mix. The app uses the same assumptions as the published Bogleheads table: 5.0% real for stocks and 1.9% real for bonds. A percentage you look up there therefore matches what the app uses. At 65 with a 60/40 mix that's 5.0%; by 80 it's 6.9%.
The spend-down age (VPW's terminal age) defaults to 100. That's deliberately past most life expectancies, because outliving your money is the risk that matters. Setting it lower means spending faster and arriving at zero sooner. It must stay above the plan's horizon (Assumptions → plan to age). At or below it, the schedule spends the whole balance while plan years remain: the taxes and Medicare due after that final 100% draw have nothing left to come from, so every path grades as failed and the success stats read zero. That's the amortization doing what you asked, not a market outcome, and the panel warns when your spend-down age does this.
Where your plan already describes an allocation (Flat or Glide), the stock/bond mix is read from it age by age, so a glide that de-risks over time lowers the withdrawal percentage automatically. On the Blended return model there are no asset weights to read, so you name the mix yourself.
Guaranteed income and the essential floor
Social Security and pensions sit underneath the rule, not inside it. VPW sets what comes out of the portfolio; your guaranteed income is added on top. That's how the Bogleheads worksheet treats it, and it's why the strategy works best alongside a solid base of guaranteed income.
Spending also never falls below the essentials you've marked fixed: Housing, Medical/Medicare, and the non-flexible share of Recurring. That floor is real protection, and it's also the honest failure mode: if the rule's number drops under your essentials, you spend the essentials anyway, and the plan can run short after all. That's not a flaw in the model. It's what running out of flexibility actually looks like.
The rule has a ceiling too: Spend at most, an offset above the expenses you entered, as a percentage or dollars a year. At 0 the rule never spends more than your expenses; at 15% it may rise that far above them in a strong market. Whatever the rule would have allowed above the cap is simply not withdrawn: it stays invested where it was, with no tax on it, which is what a household that does not need the money does. Below the cap the rule governs as before, so a bad decade still cuts. The offset follows your expenses through every phase of the plan, which is why it is an offset and not a dollar figure. Leave it blank and the rule's whole amount is spent, however large a strong market makes it; a new plan starts at 0.
You set the floor with Flexible portion of Recurring expenses, under the picker whenever a rule can act on it. Say 60% and the other 40% of your Recurring budget is essential: the guardrail may trim the 60%, and a share rule can never take you below the 40%. It is one number doing the same job from two directions: the part of your budget you could actually give up. The label names Recurring because that is its whole denominator, and the readout beside it says "essential recurring expenses" for the same reason: Housing and Health coverage are essential too, and are not in that figure.
Under My expenses, as entered the field is not shown, because nothing on that page reads it. It is still the assumption Explore's cut-back row is priced from, so it appears there instead, beside that row.
Beside the field, most this rule can reduce is what the rule PERMITS, which is not what the simulation did. Under the guardrail it is the flexible portion stopped at your floor (a $1,600 flexible portion at an 80% floor allows a $320 cut); under a share rule it is the whole flexible portion, because that rule has no floor of its own.
At 100% none of your Recurring budget is protected. Housing and Medical/Medicare stay fixed regardless, but everything else can be cut, and a bad enough run can take a share rule's spending down to those costs alone. That's a legitimate setting; just make it a choice rather than a default you never saw.
Two things the app does differently
Taxes come out on top. Here your spending number is what you spend, and the tax bill is drawn from the portfolio separately. The Bogleheads worksheet pays tax out of the withdrawal itself. So a tax-deferred-heavy plan draws more than the raw percentage suggests. That is realistic and visible in the results.
Cash counts as bonds. The published table has only stocks and bonds; if your plan holds a cash sleeve, it's priced at the bond return.
Reading the two lines under the tiles
"Spending is reduced in 38.0% of outcomes; when it is, the first time is typically at 72."
The first figure is the share of simulated outcomes in which the rule ever spent below your entered expenses. The second is the median first-reduction age among only those outcomes. The age is conditioned on a reduction having happened, because a median over every outcome would name an age at which nothing occurred whenever most outcomes never reduce.
This is why a $0 tile is not the same as "never". If 95% of outcomes make no reduction and 5% make large ones, the tile reads $0 (one outcome in ten does nothing) while this line reads 5.0%. Only when no outcome ever reduces does the line say so outright. A share too small to print at that grain is shown as "<0.1%", never rounded to zero: an outcome that happened is never reported as one that did not.
"Some expenses go unfunded in 8.0% of outcomes. Among those, the first shortfall is typically at 84."
The share is the outcomes that failed at least one obligation, and the age is the median first failure among those. It is not a spending figure and does not belong to the rule: a plan with no adjustments at all can produce a large number here, which is exactly why the two lines are separate sentences.
Both are measured from your first retirement year onward, like every other spending figure here. A working year in which the plan spends nothing is not a lean year and not a reduction.
Following the rule for real
Everything above is the simulation: what a rule would do across thousands of futures. Actually following one is a separate step, and it happens on the Overview.
Name the plan you are living under (Retirement plan, the Scenario menu, the calendar icon beside a saved plan) and the spending check-in card runs that plan's rule once a year against your real balances. It shows the budget you recorded and the budget today's balances imply, keeps them apart, and remembers what you decided.
The two sides use one piece of arithmetic, so on the same inputs they give the same answer. Under My expenses, cut back in bad years a recorded budget also seeds the projection above, which is why this panel says when it is starting somewhere other than the plan as typed.