aprise

Variance effect

Real years scatter: from 1994 to 2025 SPY’s calendar-year return ran from −36.81% in 2008 to 38.05% in 1995, with an average of 12.27% and a standard deviation of 17.86%. This page draws each year’s return at random with the average and spread you set and runs the plan thousands of times (2,000 by default). It shows the range of balances you could end up with, the share of runs in which the money lasts, and what the spread itself costs. Results are hypothetical.

How this works

One run. Each run is the planner’s month-by-month plan, except that every plan year gets its own return, drawn at random with the average and spread you set — one for the years before retirement, one for the years after.

SPY’s actual years, reshuffled (the default). Each year picks one of SPY’s 32 calendar years at random, with replacement, and shifts and stretches it to your average and spread; at the defaults it replays SPY’s own years exactly, in random order. Drawn this way a year’s standard deviation is √(31/32) of the volatility you set (17.58% at the default), and the typical run compounds at SPY’s own 10.70% — the planner’s default.

Lognormal. A smooth curve with exactly your average and standard deviation that never falls below −100%. It rarely draws a year like 2008: at SPY’s average and spread a fall of 36.81% or worse comes about once in 5,000 years, against once in SPY’s 32, so its results read a few points rosier than the default’s.

Normal. A symmetric bell with your average and standard deviation. A draw below −100% is counted as a total loss, and the page says when that happens.

Seed. The draws come from a seeded generator, so the same inputs, seed and number of runs give the same answer every time, and each run keeps its own draws whatever the number of runs. New random draw picks another seed.

Percentiles. At each age, the 10th percentile is the balance that 1 run in 10 falls below, the median the middle run, the 90th the balance only 1 in 10 beats. The bands join those points age by age; they are not the path of any one run.

Lasting. A run lasts if it pays every withdrawal in full through the age you plan until. The share of runs that last is a model result, not a probability about your future.

Variance drag. A year that loses 50% needs a 100% gain to get back, so swings cost compound growth even when the average stays the same. SPY averaged 12.27% a year but compounded at 10.70% — the gap is what year-to-year swings cost. The page shows the drag for your inputs: your average minus the typical run’s compound growth.

The spread, varied. The variance chart reruns the plan at 0× to 2× your volatility, before and after retirement together, on the first 2,000 runs with the same random draws, so the differences between its points come from the spread alone.

Sequence risk. Runs are split by how their first years of retirement went. Bad years early, while withdrawals are taken from a balance that has just fallen, do more harm than the same years later.

The planner’s spending figures. The amounts on the planner that last exactly to the end, or keep the balance’s buying power, assume every year earns the same return — no bad years. Test them here.

Why a standard deviation, not a variance. The variance is the standard deviation squared — SPY’s is 0.0319, in units of a return squared that are hard to read. The standard deviation is in the same % a year as the return, so that is what the form asks for.

What is left out. Each year is drawn independently of the last, so there is no momentum or recovery after a crash; inflation is the fixed rate you set and moves independently of returns; there are no taxes, no fees beyond those inside SPY’s returns and no change of spending after a bad year. A model result is not a forecast. Past performance does not guarantee future results.

You
Saving
Returns
Inflation

Inflation raises your spending every year and turns future dollars into today’s dollars. It does not change the return you entered: SPY’s 12.27% average is before inflation, so higher inflation means a lower real return.

Spending in retirement
Set spending as
Simulation
← Plan without variance

Carries your ages, savings and spending. Returns start from SPY’s compounded 10.70% a year, the same every year.

Every result on this page is a hypothetical illustration, not investment advice. Past performance does not guarantee future results: SPY's 1994–2025 returns and that period's inflation are assumptions here, not forecasts. Taxes, fees beyond those already inside SPY's returns, Social Security and other income are not modelled.