All the Numbers

Stress tests

From the All the Numbers wiki · The plan itself

A stress test answers one question: if this goes badly, does the plan still work? It is a way to find the weak point, not a forecast.

Each test is a copy of the plan with one change applied, run through the ordinary projection. There is no separate calculator, so every stressed run is a plan you could have typed. The result is reported as the stressed run minus the baseline.[1] The sizes are a choice, not a claim about what the worst case is.

The tests

The planner shows a row for each test that applies to the plan. The default sizes are:

  • Renewal at a higher rate. The mortgage rate is 2 points higher after 5 years. Shown only when the plan has a home.
  • Income loss. 40% less income for one year, starting 3 years in. Shown when there is pay.
  • The suite sits empty. 20 points more vacancy on any rental suite. See renting out a suite.
  • Markets do worse. Every account's return is 2 points lower, to a floor of nil.
  • Life costs more. Living costs 10% higher every month.
  • The home appreciates less. Home value grows 2 points a year less.

What each row shows

  • Net worth at the end of the plan, and the change from the baseline; see net worth.
  • Whether the plan stops. A run that cannot pay a month's bills after drawing on its accounts and borrowing to the lenders' limit is marked as breaking the plan, with the month. See borrowing when short.

What it deliberately does not

  • It does not give a probability. For the spread of outcomes under random markets, see Monte Carlo; for the effect of small changes to each assumption, see which assumption matters most.
  • It does not combine tests. Each row changes one thing.
  • The engine can also test a bad run of years in a row, and calculate early-retirement variants. The stress card does not use the former, and early retirement is shown separately.

See also

  • Expected returns and fees · The default return is 6.35% a year, nominal and net of fees, for an 80/20 stock and bond portfolio. It is an expected long-run average, not a forecast, and an account's own rate replaces it.
  • Monte Carlo: many market histories · The Simulation page runs the plan 500 times with investment returns and home prices drawn at random each year, then shows the spread of net worth and how many runs finish, borrow or run out of money.
  • Net worth, as the plan counts it · Net worth is everything the plan holds (cash, investment and registered accounts, homes at market value) minus mortgages, credit lines, debts and tax owing. It is not reduced for selling costs or for tax on money still in an RRSP.
  • Renting out a suite · Rent from part of your home is taxable income. The costs of the whole home are deductible only for the rented share, mortgage principal and land transfer tax are not deductible, and a loss offsets other income when the rental is run to earn money. A lender may count part of the rent when you apply.
  • Renting versus buying · The planner compares a plan that buys a home with the same plan that never buys, and whichever side spends less on housing in a month invests the difference at the same return. It shows when owning pulls ahead on net worth, if it does. It is a comparison under assumptions, not a verdict.
  • Scenarios and life events · The Scenarios page lets you drop dated events onto a copy of your plan, such as a job loss, a market drop or a rate jump, and see where net worth ends up. Each event is a change to the plan run through the ordinary projection, not a separate calculator.
  • When the plan runs short of cash · When a month needs more money than the plan has, it draws savings in a set order, then borrows on a home equity line at 5.13% and an unsecured line at 8.38%, as far as a lender would lend. When no lender would lend more, the projection stops.
  • Which assumption matters most · The sensitivity chart moves one assumption at a time, down and up by a fixed step, and ranks them by how much the final net worth in today's dollars changes. Inflation is often the largest bar because it moves much more than one number.

References

  1. ↑FP Canada / IQPF Projection Assumption Guidelines (sensitivity on return and inflation); Bengen (1994) for sequence-of-returns · A stress test is a pure transform of the plan run through the ordinary simulation — return shift, year-by-year return sequence (invested accounts only; savings are interest), living-costs shift, income gap, appreciation and mortgage-rate shifts — reported as shocked minus baseline; FIRE presets (full, coast, barista) are plan transforms over employment and contributions; no probabilities · effective 2026-09-19