FireMathLab

What affects your FIRE date, and when

One point of market return moves a 25-year-old's FIRE date by 8.2 years and a 55-year-old's by 0.8. The top lever changes hands 10 to 15 years out.

By Jobi·Published 4 October 2026·Updated 10 November 2026·8 min read
This guide uses the FIRE Date Sensitivity.🎚 Open the calculator

Ask what affects a FIRE date most and you get one of two confident answers. Savings rate, say the people who have read the early-retirement forums. Returns, say the people watching their fund. Both are right for somebody, and the awkward part is that the somebody changes as the plan ages. A 25-year-old's top lever is close to irrelevant at 55, and the one that barely registers at 25 is the second biggest thing on the page thirty years later.

So I measured it. The FIRE date sensitivity tool takes a plan, nudges each input both ways by a plausible error (10% for the money amounts, a full point for the real return, a quarter point for the withdrawal rate) and reports how far the date moves. One catch before any numbers: it moves a single input at a time, through a projection that assumes the same return every year, and neither of those is how a life or a market behaves. Still, the ranking is worth having. Read it as a map of where a plan is fragile, not as a forecast.

What affects one plan's FIRE date at four ages

Meera is 38. She has £120,000 invested, part of it from share options at a job she left in her early thirties, adds £1,200 a month and wants £32,000 a year in today's money. At a 4% withdrawal rate that is an £800,000 target, and at 5% real growth she reaches it at 57.9. Those are the numbers the tool opens with, so your screen already shows her chart. On screen the baseline is rounded to 58 and each bar is printed to a tenth of a year. Switch the currency selector to dollars, or euros, or rupees, and the symbols on the inputs change and nothing else, because every bar is measured in years.

Her longest bar is market returns. A point better and she finishes 1.8 years early; a point worse and she finishes 2.4 years late. That is 4.2 years between two assumptions nobody could tell apart in advance. Retirement spending comes second at 3.0 years, then the withdrawal rate at 1.9, monthly savings at 1.8 and the starting pot last at 1.3.

Now leave the plan alone and move along it. On that path she holds about £380,000 at 48 and £655,000 at 55, and typing those ages and pots into the tool returns the same finishing date, to within a tenth of a year, under a rearranged chart. For the far end, take someone of 25 with £10,000 and £600 a month chasing the same £800,000. They get there at 61.8.

Lever and nudge Age 25 Age 38 Age 48 Age 55
Market returns, 1 point 8.2 4.2 2.6 0.8
Retirement spending, 10% 3.5 3.0 3.0 3.0
Withdrawal rate, 0.25 points 2.2 1.9 1.8 1.8
Monthly savings, 10% 3.2 1.8 0.6 0.2
Starting pot, 10% 0.2 1.3 2.4 2.8

Each figure is the gap, in years, between the favourable nudge and the unfavourable one.

Read along the top row. At 25 a single point of return is worth 8.2 years, and 4.7 of them sit on the wrong side: finishing at 66.5 instead of 61.8. By 55 the same point is worth under ten months. Monthly savings follows it down, from 3.2 years to almost nothing, while the starting pot climbs from a rounding error to second place.

That is backwards from the usual advice, which says savings matter most early and returns matter later. That advice is describing something else. It is true of how the pot grows this year: at 38 Meera puts in £14,400 and growth adds about £6,300. It is not true of the date, because an error in the growth rate compounds through every year still to run, and the 25-year-old has nearly thirty-seven of them.

Spending is the odd row. It hardly moves, and the reason is that a 10% change shifts the target by £80,000 whichever age you ask at, so the bar is always answering the same question: how long does the last £80,000 take? Near the finish Meera's pot is growing by about £50,000 a year, contributions included, which makes it a year and a half from anywhere. Withdrawal rate is that same lever at a smaller nudge, since a quarter point moves the target by roughly 6%.

One plan could be a fluke, so I ran about 195,000 of them through the same engine: ages from 22 to 60, pots up to £900,000, savings from £300 to £3,000 a month, spending from £20,000 to £45,000 a year, returns between 3% and 7%, withdrawal rates between 3% and 5%, keeping the ones that reach their target by 75. Every plan with twenty or more years to run had returns on top, all 28,084 of them. No plan inside its last five years did; spending led 93% of those and the starting pot led the rest. Between ten and fifteen years out the charts split between spending and returns, and that is where the handover sits. The withdrawal rate never topped a single one.

Where the market risk goes

It would be easy to read that top row as the market mattering less as the date gets close. It does not matter less. It changes bars.

At 55, "10% smaller pot" costs Meera 1.5 years, and a 10% smaller pot is not a rare event: in the return history this site's backtest uses, shares lost a tenth or more of their real value in 24 of 124 calendar years. Type £590,000 into the current savings box and her date moves from 57.9 to 59.4. Type £460,000, a fall of about 30%, and it lands at 62.7, nearly five years late. The same 30% fall at 38, to £84,000, costs her 1.9 years. Early on, the danger is being wrong about the long-run average. Late, it is one bad year arriving at the wrong moment, which is the subject of the sequence-of-returns risk post and the thing the crash test exists to try on your own plan.

A one-point error is also the polite version of the uncertainty. This site's historical backtest carries approximate real returns for US large-cap shares from 1900, which gives 105 overlapping twenty-year stretches. Their annualised real return runs from 0.5% for the twenty years starting in 1901 to 13.3% for those starting in 1980, around a median of 6.1%, and only 24 of the 105 land within a point of that median. The UK regulator builds the same doubt into every pension illustration: the FCA's projection rules require a lower and a higher rate held three points either side of the intermediate one, with ceilings of 2%, 5% and 8% before inflation for personal pensions. Put that width on Meera's plan at 38, growth of 2% and 8% in place of her 5%, and her date runs from 53.3 to 67.8.

Fourteen and a half years, from the one input she has no say over.

Following along? The FIRE Date Sensitivity takes the numbers from here.🎚 Open the calculator

The half of the chart she can pull

Spending leads the levers Meera controls at every age in the table, and its bar still undersells it. The tool moves only the target. Cut the plan from £32,000 to £29,000 and the date comes in to 56.5, which is 1.4 years. But £3,000 a year not spent is £250 a month that can be invested, and with contributions at £1,450 as well the date is 54.8. One decision, and just over three years.

Returns cannot be bought, though charges can be cut, and inside the model they are the same thing. If Meera's 5% is what she expects after 1.2% a year in fund and platform costs, moving to 0.3% makes it 5.9%, and her date becomes 56.3. That is exactly what saving an extra £250 a month buys on its own. The fee impact calculator prices the difference for whatever you are paying now.

There is a wrinkle in the green end of that spending bar if you are in the UK. A 10% cut puts Meera's date at 56.4, and anyone under 50 today cannot normally touch a SIPP or workplace pension before 57. The tool treats her pot as one accessible pile. If most of it sits in a pension, the early date is only real once enough is held in an ISA to carry the months before the lock opens, and the pension bridge calculator sizes that gap. Other countries name the accounts differently and lock them just the same.

Bars do not add up

Each bar assumes the other four inputs stay put. Push four of Meera's against her together (a 4% return, £35,200 of spending, £1,080 a month, a £108,000 pot) and she finishes at 63.9, six years late, where the four red bars sum to 5.6. Push the same four in her favour and she gains 4.4 years against a green total of 4.7. Bad news compounds and good news overlaps.

Charts like this one draw a standard objection: they change a single factor at a time. Andrea Saltelli and colleagues surveyed the practice in Why So Many Published Sensitivity Analyses Are False and found most studies moving along single corridors that leave nearly all of the space of possible inputs unvisited. A tornado chart is exactly that design. It ranks the levers honestly and says nothing about combinations, and life deals in combinations: the market fall that shrinks the pot can arrive in the same year as the redundancy that stops the contributions.

So use the ranking for what it can carry. Count your years to go before you look at the bars. More than fifteen, and the longest one will almost certainly be returns: plan on a lower number than you hope for, then put your effort into the three that are yours to change: spending, saving, charges. Fewer than ten, and the work is different. Pin the spending figure to what actually leaves your account, and run the plan through Monte Carlo to see how it holds up across bad stretches as well as average ones. The walkthrough of the tornado chart covers the reading itself.

Then run it again next year, because the order will have changed.