Retirement spending is two decisions. Where the money comes from is the drawdown order: which account to empty first, which to leave to grow, and when each can be touched. How much to take is the withdrawal rule. Glideworth models both, and can tell you what the rule says for this year.

Drawdown order

By default a shortfall is met from cash, then taxable investments, crypto, tax-deferred accounts, health accounts, tax-free accounts and finally real assets that you have marked as available to sell. You can reorder categories or name individual accounts, draw within a category in list order or in proportion to balance, set a cash buffer to keep, and set the date or age from which each account can be touched. Statutory access ages, such as a pension’s minimum age, still apply, and an early-withdrawal penalty is applied where you set one.

Withdrawal rules

A strategy sets how much to take from the portfolio each year. The rules available are:

  • Fixed amount, inflation-adjusted or not.
  • Initial percent (the “4% rule”: a percentage of the starting portfolio, then adjusted for inflation) and fixed percent of the current portfolio.
  • 1/N, and variable percentage withdrawal (VPW), which rises with age.
  • Ratcheting, which lifts spending after strong years, with a cool-down.
  • Clyatt 95%, which never lets spending fall below 95% of the previous year.
  • Guyton-Klinger, with a guardrail and an adjustment you set (20% and 10% by default).

Each rule can have a floor and a ceiling, and you choose whether other income reduces withdrawals or lifts spending. Flexible spending is a simpler rule of the same family: when the portfolio is down by a set fraction from its high, cut flexed spending by a set fraction, in steps or linearly, on all spending or just the discretionary part.

Guardrails that watch the risk

Percent-of-portfolio rules react to the portfolio. Risk-based guardrails react to the chance of success. You set a target (80% by default) and a band around it (70% and 95%). If the chance of success leaves the band, spending is cut or raised by a step. The planner works from one shared set of simulations rather than nesting simulations inside simulations, and it turns the band into a withdrawal-rate band the engine applies year by year, so you also get a plain answer: “what do I spend this year?”

A worked illustration

The Guyton-Klinger rule in money terms, with illustrative numbers. A household starts with £1,000,000 and spends £50,000, a 5% initial rate, with a 20% guardrail. After a bad year, inflation lifts the spending to £51,000 while the portfolio has fallen to £760,000.

Value
Current withdrawal rate £51,000 ÷ £760,000 = 6.7%
Cut-above rate (5% × 1.2) 6.0%
Portfolio below which to cut £51,000 ÷ 6% = £850,000
Action Cut spending by 10% to £45,900

The tool states this the other way round as well: the portfolio level below which to cut, the level above which to raise, and the action now. Over MCP these are guyton_klinger and evaluate_guardrails; over REST, POST /api/v1/plans/{id}/insights/guyton-klinger and /insights/guardrails.

Tax strategies

Withdrawal order is also a tax decision. The planner models Roth conversions to a target bracket, an IRMAA tier, an ACA income level or a fixed income, plus withdrawal shielding and gain harvesting for US plans, and UK strategies such as drawdown order, using the personal allowance in gap years and the timing of the tax-free lump sum. The tax-strategy comparison ranks candidates (or your own) by end net worth, lifetime tax or chance of success, and a matrix compares strategies across those outcomes. For the US 72(t) early-distribution rules it compares the three SEPP methods. See solvers and tax estimates.

What it does not do

  • Rules model your intentions, not your behaviour. A rule says what a disciplined household would do; real households may not.
  • No dedicated go-go and slow-go phases. You can approximate them with flexible spending and milestones.
  • The tax optimiser has limited depth. It ranks candidates it generates (or yours); it has no beam search or depth setting.
  • No Flex versus No-Flex pulse table. Risk-based guardrails are compared instead.
  • Strategy and income-by-rate heatmaps are partial: they run as background analysis jobs.
  • Survivor benefits are not added automatically in survivor what-ifs.
  • Not advice. A strategy is a model of a rule, and tax results are estimates.

Go further

  • Monte Carlo for the simulations guardrails read.
  • Explanations: “why did this withdrawal come from that account?”
  • Branches: compare two strategies side by side.

See where your plan lands.

Free to self-host today, with every feature. The hosted service opens later, at $9 a month or $90 a year.