How spending guardrails are calculated

Risk-based guardrails need the answer to one question, many times over: at this balance and this spending, what's the chance of needing a cut later? A full simulation for every possible balance would be too slow on a phone. The app answers it with a precomputed lookup table.

The rule

• Spending level. Spend at the level where the chance of a later cut equals your comfort level: about 1 in 10, 1 in 5 or 3 in 10.
• Lower guardrail. The investment balance at which the chance of a cut, at your current spending, reaches 75%.
• Upper guardrail. The balance at which that chance falls to zero.
• When one is crossed, spending goes back to the comfortable level for the new balance. The change is at least 5% of spending, but never past the comfortable level.
• Between the guardrails, spending stays where it is.

An earlier version cut only 10% of the gap between current and comfortable spending, while raising the whole gap. On sample plans, that lopsided rule lowered the chance the money lasts, so cuts and raises are now symmetrical.

Limits

• Set-aside. By default, 10% of investments, rounded to $1,000, is left out of what the guardrails count on and added back when guardrail balances are shown. You can change it or set it to zero. For working households it applies from retirement.
• Ceiling. An optional maximum monthly spend. The starting level and any raise stay at or below it.
• Paid first. Healthcare, planned one-time purchases and debt payments are set against guaranteed income before the guardrails are worked out.

How retirees are modeled

For a retired household, the engine treats spending as a level, inflation-adjusted withdrawal:

1. Spending plus its share of taxes is the total need. The tax share comes from a full projected year, capped at 60%.
2. Guaranteed income, such as Social Security and pensions, covers part of it. Future guaranteed income is valued at a 1% real discount rate.
3. The rest is a level withdrawal from investments, after the set-aside.

The risk surface

The lookup table is built once per plan:

1. Simulate 1,000 paths of inflation-adjusted returns, up to 70 years long, using your plan's simulation method and its long-run asset mix.
2. For each path and each horizon, record the highest level withdrawal rate the path could have sustained.
3. Sort them. For any withdrawal rate and horizon, the share of paths that sustained it is the chance of success.

Looking up the table takes microseconds, so the guardrails can be recomputed every time a balance changes.

Working households

For people still working, pay and savings matter too much for the level model. The app runs the whole plan instead. It finds the balances at which today's spending's chance of a cut would reach 75% and 0%, by applying market moves to your investments and simulating each one.

How accurate it is

The tolerance was set before measuring: within 6 percentage points of a full simulation for a retiree, and within 3 points of a direct simulation of level withdrawals. Measured results:

• Portfolio only: largest error 1.4 points.
• Full engine, a retiree with only Roth savings spending $45,000 a year from $1,000,000, 1,000 futures: 74.1% from the full engine, 74.4% from the lookup table.

Larger errors come from three places, and the screen notes them next to the guardrail figures:

• taxes, which the table folds in through a ratio
• guaranteed income, valued at a fixed real rate
• asset mixes that change over time, where the table uses the long-run mix

Chance of a spending change

The chance of a cut or a raise in coming years comes from 500 full-engine futures in which the household follows the guardrail rule. Retirees start from their current spending. Working households spend their current amount until retirement.

These guardrails are the app's own method, inspired by published research. They're estimates, not guarantees or advice.