Returns are not a straight line

Every projection on this site draws a smooth curve. So does every other investing calculator on the internet. That curve will not happen, and it is worth understanding exactly how it will fail to happen, because the failure modes are different depending on what you are doing.

What "averages 7%" hides

A market that averages 7% over a decade does not deliver 7% in any particular year. It delivers something like +18%, −9%, +23%, +2%, −14%, +31% and so on. The average is a summary of the outcome, not a description of the ride.

While you are only paying money in, that mostly does not matter. Your contributions during the bad years buy more units at lower prices, the good years lift everything, and the end result lands reasonably close to what the smooth curve suggested. Volatility is uncomfortable but not especially costly.

Then you start withdrawing, and order starts to matter enormously

Consider two people with identical pots, identical withdrawals, and — this is the important part — identical average returns over thirty years. The only difference is the order the returns arrive in. One gets the bad years early; the other gets them late.

The one who gets bad years early is selling units to fund their income while prices are down. Those units are gone permanently, and they are not there to participate in the recovery. The one who gets bad years late has already banked a decade of growth before the fall arrives, and is drawing from a much larger base.

Two retirements with the same average return, the same starting pot and the same spending can end decades apart. The only variable is the order.

This is called sequence-of-returns risk, and it is the single largest thing a smooth-curve calculator cannot show you. It is why the drawdown calculator carries a warning on the page rather than in a footnote.

Why this site shows three numbers instead of one

We could have built a Monte Carlo simulation — run the plan ten thousand times with randomised returns and report the distribution. Plenty of tools do. The reason this site does not is that a Monte Carlo requires you to assume a volatility figure and a distribution shape, and those assumptions are every bit as unknowable as the return itself. The output looks far more scientific than its inputs justify, and "your plan succeeded in 87% of simulations" is a precise-sounding number resting on guesses nobody labels.

So the approach here is deliberately cruder and more honest: run your plan at three different assumed returns, put them next to each other, and let the width of the gap do the talking. When the cautious and optimistic answers differ by a factor of two — and over thirty years they usually do — that gap is the actual finding.

What to do with a projection

• Use it comparatively, not predictively. "Does five more years help more than an extra 100 a month?" is a question the arithmetic answers well. "How much will I have in 2056?" is not.
• Plan from the cautious row. A plan that only works at the optimistic assumption fails quietly and reveals itself far too late to fix.
• Re-run it. A projection is a snapshot of assumptions, and yours will change. It is worth five minutes once a year, and nothing in between.
• Distrust precision. Any tool reporting your 2056 balance to the nearest unit is communicating a confidence that nothing about the underlying problem supports.

The honest summary

The arithmetic on this site is exact and tested. The inputs are guesses. Multiplying an exact calculation by a guess gives you a guess, dressed up in the visual language of precision — and the dressing-up is the part worth being alert to, here as much as anywhere else.