TL;DR. A spending-plan rule decides what your base annual spending is in any given year of the simulation. The seven rules trade off lifestyle stability against portfolio responsiveness: stable rules keep your standard of living smooth but can sprint into a depleted portfolio in bad sequences; responsive rules absorb market shocks by cutting spending, which fattens the failure tail at the cost of more year-to-year variation.

Rule comparison

Rule	One-line behavior	When it shines	When it bites
Non-Inflation Adjusted	Same nominal dollar amount every year.	Short horizons; conservative back-of-envelope.	Long horizons. Inflation silently halves your real lifestyle.
Inflation Adjusted (CPI)	Initial spending grown by cumulative CPI each year.	The classic "4% rule" baseline; predictable real lifestyle.	No feedback from the portfolio. Keeps spending through bad sequences.
Percent of Portfolio	A fixed percentage of the current balance, every year.	Cannot run out by construction; tracks real returns closely.	Lifestyle is volatile. A 30% market drop is a 30% pay cut.
Variable (Z-Value)	Blends inflation-adjusted spending with a portfolio-ratio nudge.	Mild responsiveness without large lifestyle swings.	Tuning z takes thought; defaults are a starting point.
VPW	Amortized withdrawal: spread current balance over remaining years at an assumed return.	Mathematically targets a chosen ending value; principled.	Spending grows late in life as denominator shrinks; needs a floor for comfort.
Hebeler Autopilot	Weighted blend of inflation-adjusted spending and an RMD-style draw (balance divided by an IRS life-expectancy factor for your age).	Actuarial by construction: spending tracks both the portfolio and your remaining life expectancy, so it rises naturally with age.	Late-life spending climbs as the life-expectancy factor shrinks; the RMD half swings with the market, so a floor is worth setting.
CAPE	Spending rate keys off Shiller's CAPE-derived earnings yield.	Spend more when stocks are cheap, less when they are expensive.	Sensitive to assumed CAPE values; behaviour leans on long-term mean reversion.
Guyton-Klinger	Inflation adjust each year, then cut by 10% if your withdrawal rate has spiked, or raise by 10% if it has fallen.	Strong failure-tail protection while staying mostly inflation-stable.	The "guardrail" cuts can hurt in early-retirement sequence-of-returns risk.
Risk-Based Guardrails	Each year, size spending to a target chance of success: the share of historical scenarios in which your plan survives. Raise promptly when over-funded; cut reluctantly when risk rises.	Steers directly at the metric you care about (success), with asymmetric raise/cut behaviour tuned to a historical engine.	Much slower; it runs hundreds of sub-simulations per year to measure success.

Risk-Based Guardrails (chance-of-success targeting)

Most rules above watch a proxy for risk: your withdrawal rate, the CAPE ratio, your portfolio vs. its starting value. Risk-Based Guardrails watches the thing you actually care about: the chance of success itself.

What "chance of success" means here. Each year the engine runs a fan-out of forward scenarios from your current portfolio and counts the share in which your plan would have lasted your full plan length. That share (a frequency over history, not a crystal-ball probability) is your chance of success. "80%" means your spending is set so the plan survived in about 80% of those scenarios.

How the guardrails move spending (raise promptly, cut reluctantly).

• Raise: when success climbs to your raise threshold (default 100%, i.e. it would have survived everything), you're under-spending, so spending steps up toward the target.
• Cut: you only tighten up when success falls to your cut threshold (default 25%). The cut then lowers spending just until success climbs back to your recovery level (default 45%), set between the cut threshold and the target. Because more safety always means less spending, a higher recovery means a deeper cut (lower spending, more safety) and a lower recovery means a gentler cut (more spending, but nearer the edge). It deliberately stops short of the target, so a cut trims reluctantly instead of slashing all the way back.
• Hold: in between, spending just keeps pace with inflation.

The horizon shrinks as you age (a 4% draw is safer with 12 years left than 30), which is what lets the guardrails ratchet spending up over a successful retirement. Horizon comes from your Simulation Duration in Time Settings. There's no separate longevity input.

How it stays fast. Sizing spending to a target success rate could mean running hundreds of sub-simulations inside every year of every scenario, a simulation within a simulation. Instead, the engine builds a single chance-of-success model once: it relates how well-funded you are (your assets vs. the present value of your remaining spending, net of income like Social Security) and your remaining horizon to the chance of success, then looks each year up against that model across the full sweep of historical cycles. That makes a Risk-Based Guardrails run a quick background analysis with a progress bar. Because the model is built from your starting portfolio, it's a close estimate rather than an exhaustive re-simulation: accurate for typical plans, with a little more give when your account mix shifts a lot late in retirement.

What the progress bar is doing. Before it can count scenarios, the run has two setup steps, and the panel names each one: Preparing market scenarios while it builds the paths, then Calibrating your guardrails while it builds the chance-of-success model. On a historical backtest these pass quickly. Combined with Monte Carlo returns they are most of the run — the guardrails are re-solved for every iteration — so the counter can take a few minutes to start moving. That is why 1,000 iterations is usually the right choice for this plan: higher counts mostly add setup time.

Cached results. A completed Risk-Based Guardrails analysis is saved and shown again instantly when you come back; it only re-runs when something that feeds the simulation changes (accounts, spending, people, glide path, custom market series, and so on). If you edit your inputs while looking at an older result, the year-by-year deep-dive will ask you to re-run the analysis so the details always match your current plan. That includes a working-years period: changing what you spend before retirement changes the portfolio the solver starts from, so it starts a fresh analysis rather than showing you the old one.

Guardrails applies from retirement, not from today. It has nothing to solve while you are still earning - the question it answers is how much you can safely draw down. So if your retirement year is still ahead, your working-years spending comes from the time period in front of the Guardrails one: press Add a working-years period on the Base Living Expenses editor and you get a period covering the years before you retire, then Guardrails from the year you retire. Guardrails is only ever valid as the last period, and it has to start exactly at your retirement year.

Spending Floor and Spending Ceiling

The optional Spending Floor and Spending Ceiling are guardrails that clamp whatever the spending rule produces in a given year. They are entered in today's dollars; each year the engine multiplies them by cumulative inflation before applying them, so a $60,000 floor keeps the same real purchasing power across the whole simulation.

In the editor the two sit on a range track with your yearly spending: a rail with a sky marker for the floor, a purple one for the spending amount and an emerald one for the ceiling, over three cells labelled Floor, Yearly spending and Ceiling. Position on the rail is the label, so a floor dragged above a ceiling is visible before Save turns grey. Percent of Portfolio, VPW and CAPE compute their own spending. On these rules, the middle marker is hollow or absent, because that number is the engine's, not yours.

Under each form is a worked example: one line of arithmetic from your own numbers, showing what the rule would spend after a specific market move and which limit held. When the three amounts are out of order, that same line carries the warning instead.

The mechanic is simple. Every rule above computes a candidate spending value for the year, and then the engine clamps it:

• If the candidate is below the floor (after inflation), use the floor.
• If the candidate is above the ceiling (after inflation), use the ceiling.
• Otherwise, use the candidate as-is.

What each rule's own controls are called

The values themselves have not changed. The editor uses new names and a new format for how percentages are typed.

• Variable Spending: Market response, with Steady, Balanced and Responsive presets for 0.25, 0.5 and 0.75 (the old Z Value).
• Guyton-Klinger: two sentences, Prosperity (up more than X%, raise spending Y%) and Preservation (down more than X%, cut spending Y%). These four are typed as whole percents now. Enter 20, not 0.2.
• Hebeler Autopilot: a two-color bar you drag to split spending between the Steady part (CPI) and the Balance part (RMD). The two parts always add to 100%. Use age of chooses whose age sizes the balance-based part, while the spending period controls when the rule applies. For a couple of different ages, choose either person; the period's year selector shows both ages. If the selected person dies or is removed, the rule uses the first living person's age. Older saved plans keep their original age-based timing until you choose a person.
• Percent of Portfolio: Spend each year X% of portfolio, with an estimate of what that is against your accounts today.
• VPW: Leave behind (the old Future Value) and Assumed return.
• CAPE: the formula itself, Rate = X × CAPE yield + Y%. The add-on is typed as a percent now. Enter 1.0, not 0.01.
• Risk-Based Guardrails: unchanged controls, with the four guardrail markers still dragged on their own band.

Clamping happens on every plan, including CPI and Non-Inflation Adjusted, but it only matters when the rule produces values that drift off the starting amount. That mostly happens with the variable rules.

The floor has to sit at or below the ceiling, and the editor blocks a save while it does not. The clamp applies the floor last, so a floor above the ceiling would lift spending back over the limit you set and the ceiling would do nothing at all.

Why they matter for variable plans

Variable rules (Percent of Portfolio, VPW, CAPE, Hebeler, Guyton-Klinger, Z-Value) tie spending to the portfolio. That responsiveness is the point. It's also what makes those rules produce spending numbers you would not actually live with. A Percent-of-Portfolio plan at 4% on a $1M portfolio says "spend $40k"; if markets drop 50% the next year, the same rule says "spend $20k". A floor at $30k stops the cut at $30k. A bull run that doubles the portfolio suggests "spend $80k"; a ceiling at $55k stops the climb there.

Time periods

Everything above applies one rule to your whole plan. Most people do not actually spend that way. The Base Living Expenses editor lets you split the plan into time periods, each with its own rule, its own amount, and its own floor and ceiling. Press Add time period to split the last one in two. The usual shape is the "retirement smile": more in the early, active years, less in the slower middle years, then a lower steady number later on.

A time period is bounded by a calendar year, and the editor shows everybody's age alongside each year so you can pick the boundary by age instead. The first period starts at the first year of your plan; every later period starts the year after the one before it ends, and the last period runs to the end. That is why you only ever set a "through year" and never a start year: gaps and overlaps are not possible, so there is no error state to clear. Move one boundary and the next period follows it.

What carries across a boundary, and what does not

• Last year's spending carries. The first year of a new period sees what you actually spent in the last year of the previous one. This only matters for the rules that build on last year's number, Guyton-Klinger above all: it will raise, cut, or hold from the real previous year rather than restarting from its own starting amount.
• Everything else starts fresh. A time period's rule runs from the numbers you typed into that period, not from anything the previous period accumulated. If a later period is Guyton-Klinger, its guardrails are measured against the values on that period's own form.
• Each period clamps itself. The floor and ceiling belong to the period, not to the plan. A lower floor in a later period is the whole point: you can accept a deeper cut at 80 than you would at 60.

The chart

Switch to Chart for a read-only picture of your spending over time. Up to three lines: Base Living Expenses in the middle, with the Spending Ceiling above it and the Spending Floor below. The lines are drawn as a staircase rather than a slope, because that is what the plan actually does: each period holds its number flat and steps at the boundary.

The chart does not edit anything. Hover a line to see its amount for that year, and the period under the cursor is highlighted in the table below. Click a year to pin that period's rule and amounts. Every change, to an amount, a rule or a boundary, is made in the List view.

Some periods show fewer than three lines, and that is deliberate. Inflation Adjusted and Non-Inflation Adjusted spending is fully determined by the amount you typed, so a floor or ceiling could never come into play and none is drawn. Percent of Portfolio, VPW and CAPE work the other way: they compute their own yearly amount from your portfolio, so they have a floor and a ceiling but no base line. Variable, Guyton-Klinger and Hebeler have all three.

One caveat the chart flags with "(starting point)": Guyton-Klinger uses its amount once, to set the withdrawal rate it starts from. After that it raises and cuts from last year's actual spending, so its base line is where the rule begins rather than a prediction of what it will spend.

A worked example

An early retiree who stops working at 45 and expects three distinct stretches:

• Through age 52. $60,000/yr, floor $45,000. The travel years.
• Through age 70. $48,000/yr, floor $38,000. Settled down, kids launched.
• Age 71 and after. $40,000/yr, floor $30,000. Lower discretionary spending, and a floor that reflects what is genuinely fixed at that age.

Run that and the year-by-year spending chart shows two clean level shifts, at 53 and at 71. Because these are real base spending numbers rather than expense line items stacked on top, everything downstream sees them: withdrawal sizing, bracket fill, guardrail triggers, and the success rate itself.

A second shape, for someone still working. Say you retire in five years and your youngest finishes college in three:

• This year through two years out. $110,000/yr. Two kids at home, one in college.
• The following two years. $85,000/yr. College done, still working.
• Retirement onward. $70,000/yr, Percent of Portfolio at 4% with a $55,000 floor.

The first two periods land entirely in your working years, which is the point: one number for the whole run-up cannot say that college ends. Those years still draw on income first and only touch the portfolio for the shortfall.

Building the number from a budget

Every rule above starts from one figure: what you spend in a year. Most people do not know that number to the dollar, and guessing it low is the single most expensive mistake you can make in a plan. Base spending drives withdrawal sizing, bracket fill, guardrail triggers, the funded ratio and the success rate. Build from a budget, under the yearly amount on any period, turns that one figure into a worksheet.

Starting points

Nobody should face a blank page, so the worksheet opens on a picker with four ways in:

• Blank. Includes every category with no amounts.
• Split my current amount. Spreads the number already on your plan across a typical household's categories. The total does not change, so linking the budget cannot silently move your plan; you are just checking a number you already had.
• Typical US household and Retired household, age 65 and older. These use real dollars from the Bureau of Labor Statistics Consumer Expenditure Survey (Table 1300, 2024). The worksheet names the table and year under each one. Offered on US plans only.

Once you pick one, the picker folds away to a Start over from a template link so the worksheet has the room. Starting over replaces every line in the budget, and the worksheet says so before you choose.

One budget per time period

Every time period can carry its own budget, and they are independent: a working-years budget with commuting and a mortgage in it, and a retirement budget with travel and Medicare, each feeding its own period's yearly amount. Build one from the Build from a budget link under that period's amount, exactly as you did for the first.

When another period already has one, the picker offers a Copy from card for it, so a second budget starts as a copy of the first rather than 38 blank lines. It is a copy and not a link: changing a line in one never touches the other. Deleting a time period deletes its budget along with it.

Lines, frequency and the Must Spend tag

Each line carries a name, an amount, a frequency (per month, per quarter or per year) and a Must spend tag. Changing a line's frequency converts its amount, so $12,000 per year becomes $1,000 per month and the totals do not move. Amounts are in today's dollars; your plan's own inflation handling takes it from there, exactly as it does with a typed number. The totals bar keeps two running figures: Like to spend, every line, and Must spend, the lines you tagged essential. Click the Must or Like chip with the swap arrows at the start of a line to switch its tag: Must spend lines are blue, the rest are purple, and a line your plan already models elsewhere turns amber and is left out of both totals.

A line is not the place for a cost with its own start and end year. Examples include a car loan that ends in six years, a child's tuition, a mortgage that gets paid off. Those are events, and the expense types page has the table for deciding which is which.

What the two totals feed

With the budget linked, Like to spend becomes the period's yearly amount. The field turns read-only and says where the number came from; typing your own amount again unlinks it, and the worksheet is kept either way.

Must spend has two homes, and which one you get depends on your rule:

• On a rule the engine clamps against a floor (Variable, Guyton-Klinger, Hebeler, Percent of Portfolio, VPW, CAPE, Risk-Based Guardrails), Use Must Spend as the floor starts checked for a new budget when the floor is available. You can uncheck it, and saved budgets keep your choice. A bad market can then trim the extras but never the essentials. Percent of Portfolio, VPW and CAPE have no base amount, so the floor is the only thing a budget can set on them: the switch starts on, and with it off the button reads Save budget because the worksheet is kept but the plan does not change.
• On Inflation Adjusted and Not Inflation Adjusted the floor is inert, so instead you get Run at Must Spend: it opens What-Ifs on the Spending Flex row with the cut already filled in, and the comparison panel shows the two side by side. A spending What-If stops at −50%, so a Must Spend total less than half of Like to Spend runs at −50% and the worksheet says so.

Must spend, the plan, and the ceiling

Those three numbers are one story. Must spend is the line your spending never drops below. Like to spend is the plan: what you spend in an ordinary year, and what the budget writes into the yearly amount. The Spending Ceiling is how much a good market is allowed to raise it, and it stays your number. The worksheet never asks what a great year looks like, so it never answers for you.

Because a ceiling below your budget is not a plan the simulation can run, linking a budget raises a ceiling that would sit at or below the new yearly amount to 10% above it, and leaves a ceiling that already clears your budget exactly where you put it. If every line is tagged Must, there is nothing between the floor and the plan, so the floor tick box switches off and says why.

All three only do their work on a rule that can move spending year to year. On Inflation Adjusted the floor and ceiling are inert, so the worksheet offers to switch that period to Variable Spending (the rule comparison at the top of this page has the full trade-off), where your essentials hold the bottom and good markets pay for the extras.

Lines your plan already models

Some budget costs are already modeled as events. These include rent or a mortgage payment, property tax, home insurance and maintenance on a property you own, health premiums, loan payments, charity, tuition. Each matching line is struck through, explained, and left out of both totals, with a link to the event in question. Counting it twice is the mistake an itemized budget invites, and it is the one that quietly makes a plan look worse than it is.

Estimate from my income

If you cannot itemize but you do know your salary, your savings rate and your mortgage, the worksheet can work backwards: take-home pay from each job, plus other income, less what your plan already saves and already models as an expense. What is left is what you have to spend. You can drop it on "Everything else" or spread it across the typical shares.

It is a starting point and nothing more. Taxes are estimated at single-filer rates on this year's figures, and anything that starts later is not counted. If your modeled income does not cover your savings and expenses, it says that rather than showing a zero.

Reading it back

In the year-by-year results, the Base Spending row gains a By budget category disclosure that splits the year's figure by each line's share. That is a view of the budget as you have it saved now, applied to a number the engine produced. It is not a record of what the simulation spent category by category, and the caption says so.