Steady pay or a pot that cannot die
Once you are drawing from a portfolio you must choose between two things you cannot have together: a predictable income and a guarantee the money lasts. Every withdrawal strategy is a position on that spectrum. Take a fixed amount regardless of markets and your income is stable but the pot can be exhausted. Take a percentage of whatever the pot is worth and it can never run dry, but your income falls exactly when markets do, which is often exactly when you least want it to. Everything else sits in between.
Three positions on the spectrum
Fixed real is the default the Trinity study tested: take the same inflation-adjusted amount every year and ignore markets entirely. Maximum predictability, and the only one of the three that can genuinely fail.
VPW (variable percentage) takes a set percentage of the current balance each year. Arithmetically it cannot reach zero, because you always take a share of what remains. The cost is volatility in your own spending: a 30% market fall means a 30% pay cut, in the year it happens. That cut has to land somewhere, and every variable strategy quietly assumes you can absorb it. If most of your budget is fixed costs, VPW's worst year is not survivable and the comparison below will mislead you.
Guyton-Klinger guardrails start with a fixed withdrawal, but watch what it represents as a percentage of the current pot. If a market fall pushes it above an upper guardrail, cut spending; if growth pushes it below a lower one, give yourself a raise. Inside the band nothing happens and each adjustment is 10%, which is where the rule gets its reputation for leaving your income alone. Measured against every historical start year that reputation is generous: a forty-year retirement collects around twenty adjustments, close to half its years, and about three of them are raises for every cut. Checking annually is still plenty, since nothing can trigger between checks. Bear in mind too that the published bands are conventions, not laws: different parameters produce different results, and the rules came from a specific US study. What survives all that is a rule most retirees converge on without knowing it has a name, because it matches how people actually behave, deciding the size of a correction in advance rather than in the middle of a bad year.
Ingrid's worst year
Ingrid retires at 55 with £700,000 and wants around £28,000 a year (exactly 4%). Running each strategy through the same thousand simulated markets:
| Strategy | Success | Median income | Worst-year income (p10) |
|---|---|---|---|
| Fixed real | 61% | £28,000 | £6,700 |
| VPW | 100% | £27,000 | £5,700 |
| Guardrails | 99.9% | £26,300 | £5,800 |
Each column hides something, so read them together. Fixed pays exactly £28,000 every year right up until it cannot: nearly four runs in ten exhaust the pot, typically in her early eighties, ages the ONS's life expectancy data gives a 55-year-old every chance of seeing, and the £6,700 worst-year figure is the sound of a failing run scraping its last balance. VPW cannot fail, yet its median income drifts slightly below the fixed £28,000, because 4% of a volatile pot means 4% of a median path that grows more slowly than its average, and its worst tenth of years pays £5,700 against bills sized for £28,000. Guardrails survive all but one run in a thousand with a similar tail, the difference being that its cuts arrive as deliberate 10% steps rather than whatever the market dictates that year. The honest summary is starker than the usual telling: flexibility is not a free lunch. It converts the risk of running out entirely into the certainty of some very lean years. How much of that variability you could genuinely live with is the question underneath all three, and the strategies guide comes at the comparison from that direction. For what the guardrail rule does to a real retirement rather than an average one, the post on guardrails versus fixed withdrawals follows the 1966 cohort year by year.
Which position suits you comes down to the gap between essential and target spending. If your essential spending is well below your target income, VPW or guardrails are strong. If nearly all of it is committed, fixed with a lower rate and a cash buffer is safer, and the withdrawal rate explorer shows what that rate should be. In practice most retirees mix the approaches anyway: cover essentials with guaranteed income like a state pension, then run something variable on the discretionary portion. The result is a floor with an adjustable layer on top.
A few things flatter the models here. Real flexibility is lumpy: a simulation cuts spending smoothly by percentages, while households cut by cancelling a holiday or delaying a car, so the direction is right but the granularity is not. Tax is not modelled, and it interacts with variable income in ways that can favour smoothing. And the usual simulation caveats apply, as in Monte Carlo.
One consequence people miss until late: the strategy changes how much you need to retire, often more than expected. A flexible strategy can support a higher initial rate, which lowers the target pot, worth testing against your FIRE date before assuming you need another year.