The idea is simple enough to hold in your head. There is some balance which, if you never added another dollar to it, would grow on its own into a retirement you can actually afford. Reach it and the arithmetic changes character: you still have to earn a living, but you no longer have to save. You can take the job that pays less. You can go to four days a week. You can stop opening the statement.

The number is worth knowing precisely, because people make real decisions with it. And the formula everybody uses to find it is wrong in a specific and consistent direction: it tells you the number is much bigger than it is.

The formula everyone uses

Every Coast FIRE calculator on the internet does the same two steps.

First, your FI number (what you need on the day you retire) is your annual spending divided by a safe withdrawal rate, conventionally 4%. That is the familiar 25x.

It is worth being clear which way that dial turns, because it reads backwards to almost everyone the first time. The withdrawal rate is not how much you spend. Your spending is whatever you decided it was. It is what fraction of the portfolio that spending represents. So a more cautious 3.5% is 29x rather than 25x, and it asks for a bigger pot, not a smaller one, because the same $70,000 has to come out of a portfolio you are willing to draw on more lightly.

Then discount it back to today at your expected real return:

Coast number = FI number / (1 + r)^years

Take a 42-year-old who wants $70,000 a year and plans to stop at 65. Their FI number is $1,750,000. At 7% growth and 2.5% inflation the real return is 4.4%, and over 23 years that discounts to $651,418.

So: $651,418 today, and you never have to save again. It is a clean piece of arithmetic and it has the great virtue of fitting on a napkin.

It is also describing a retirement that almost nobody has.

It assumes your portfolio pays for everything

Look at what 25x actually says. It says your portfolio must produce every dollar you spend, for the rest of your life, on its own.

For most people it does not have to. Social Security is going to pay something. Many people have a pension. Some have rental income. Those are not rounding errors. For a household spending $70,000 a year, a $30,000 Social Security benefit is nearly half the bill, arriving every year, inflation-adjusted, for as long as you live. The formula has no way to express it, so it silently asks you to save as though it did not exist.

There is a second, quieter assumption. The 4% rule is built to survive indefinitely: it is designed so the money never runs out, no matter how long you live. A projection that runs to 100 is answering a narrower question, and a narrower question needs less money.

The Coast FIRE tab on this site does not use the formula. It takes the plan you have already described (your accounts, your spending, your Social Security timing, your tax rates) switches every contribution off, yours and your employer’s, and runs the whole projection forward. Then it solves for the balance that just carries it. Same household, same day, both numbers:

The gap is $284,269. It is not a rounding difference or a modelling quibble: it is 44% of the number, and it is the difference between “I am nowhere near” and “I am closer than I thought”.

It comes from the two assumptions, and they can be separated. Run the same household with Social Security deleted and the projection asks for $547,646:

So about $104,000 of the gap is the 4% rule aiming at perpetuity where the projection aims at age 100, and about $180,000 of it is Social Security. The second is the larger effect, and it is the one the napkin cannot ever capture, because there is nowhere in FI / (1 + r)^n to put it.

Where the line actually is

The number on its own is a milestone with no date attached. The more useful question is when you get there, and that has a shape worth looking at.

Two lines. One is your balance if you carry on contributing exactly as you are. The other is what you would need at each age to coast from that age, which rises, because every year you wait is a year of compounding you no longer get. Where they meet is your Coast FIRE date.

Nine years, not never. That is the thing the chart tells you that the number alone does not: this saver is not choosing between saving until 65 and giving up. They are nine years from the point where continuing to save becomes optional.

Notice also how the two lines behave after they cross. The requirement keeps rising, but the balance rises faster, because it is compounding and being fed. Past the crossing, every additional year of contributions is buying margin: an earlier retirement, a bigger number, a worse decade absorbed, rather than the plan itself.

What this is assuming

Coast FIRE is a projection, and it inherits every weakness projections have. Four are worth stating plainly.

The spending target is doing most of the work, and it is the number most likely to be wrong. Everything above is downstream of $70,000 a year. Expenses at 42 rarely resemble expenses at 65. Children arrive and leave, parents need help, houses need roofs. If the real figure turns out to be $85,000, the coast number moves by more than any assumption argued about on the internet.

A flat 7% hides the order the returns arrive in. The projection grows the balance by the same percentage every year. Real markets do not, and a bad first decade of coasting cannot be made up by a good second one in the way the average implies. The Monte Carlo tab exists for exactly this reason; run the plan through it before you act on a single deterministic line.

Stopping early means paying for health insurance yourself. Anyone who coasts out of a job with benefits before 65 is buying their own cover until Medicare, and that is a real annual cost this projection does not model as a separate item. If that is the plan, put it in the spending target.

Coasting is easier to start than to stop. This one is not arithmetic. The habit of saving is difficult to restart once broken, and the flexibility you bought by coasting can quietly become a floor you cannot climb off. Reaching the number is permission to stop, not an instruction to.

And one thing that is not a caveat but is often mistaken for one: the projection lets the portfolio run down toward zero by 100. That is deliberate, since money left over at 100 is money that was not spent, but it is a different target from the 4% rule’s, and it is part of why the numbers differ.

What to do with this

Put your own figures into the calculator, real balances and real contributions, your actual Social Security estimate from ssa.gov rather than a guess, and open the Coast FIRE tab. It will show you both numbers, the gap between them, and the age the lines cross.

Then do the thing the number is actually for, which is not to stop saving. It is to find out how much of your current saving is load-bearing. Most people discover they are further along than the napkin told them, and that some part of what they are putting away is buying margin rather than buying the retirement. Knowing which is which is what lets you make the trade deliberately (fewer hours, a better job, a worse-paid one that you like) instead of saving hard until 65 because nobody ever told you when you could stop.

If the compounding underneath all of this is unfamiliar, the ten years that beat thirty makes the same point from the other end: it is time, not the rate, that does the work.

It is the wrong thing to argue about, or at least the wrong thing to argue about first. The input that does the most work in any of these projections is not the rate. It is how long the money is left alone.

Here is that claim with numbers attached. Everything below came out of the calculator on this site, one run at a time, with nothing changed between runs except a birthday.

Same $6,000, three starting lines

One Roth IRA. $6,000 a year, which is $500 a month. 7% a year. Stop at 60.

The only difference is when the first payment goes in.

The twenty-year-old pays in twice what the forty-year-old does and ends up with nearly five times as much. That gap is not a reward for being twice as diligent. It is what the extra twenty years did to the same deposits.

The last decade does most of the work

Look at the first chart again. For most of the twenties the line is flat enough to be discouraging. After ten years of paying in, the account holds $82,899, and $60,000 of that is money that was simply handed over. Nothing much appears to be happening.

Then look at where the balance actually arrives.

At 7% a balance doubles a little over every ten years. So the years you add at the front of a plan are not worth what they look like at the time; they are worth whatever the last doubling turns them into. Run the numbers on that first chart and the final ten years, from 50 to 60, add $631,046, which is more than the first thirty years produced in total.

That is the whole argument, and it cuts both ways. The years are cheap when you have them and unbuyable once you don’t.

The ten years that beat thirty

There is an old illustration that gets passed around, usually without anyone checking it. Someone saves hard through their twenties, stops completely at 30, and never adds another dollar. Someone else starts at 30 and pays in steadily for the next thirty years. The claim is that the first one wins.

They do.

Getting this out of the calculator takes two passes, because an account here carries one contribution figure that runs until you retire. There is no “stop paying in at 30” field, and inventing one would complicate every other plan to serve this one story. So: run it to 30, read the balance, then start a second plan from that balance with the contribution set to zero.

So the saver who paid $60,000 into the account between 20 and 30 and then stopped has $631,046 at 60.

The saver who started at 30 and paid in every year until they retired put in $180,000, three times as much, across three decades of automatic transfers, and has $566,765.

Three times the money, and they finish $64,281 behind.

To actually match the early saver, the late one has to find $6,681 a year rather than $6,000. Not a dramatic difference in the monthly figure, and $200,415 in total: three and a third times what the first saver ever paid in, to end up in the same place.

What this is assuming

Everything above is one deterministic run of the same arithmetic the app applies to every account: the balance earns its return, then the year’s contribution lands on top, thirty or forty times over. Which means:

• 7% every single year, with no variation. Real markets do not do this, and the order in which good and bad years arrive matters a great deal, and that is what the Monte Carlo tab is for. A straight line is a teaching aid.
• These are future dollars, not today’s. The Growth chart is not inflation-adjusted. $1,197,811 in 2066 does not buy what $1,197,811 buys now, and at 2.5% inflation it is closer to $450,000 in today’s money. The comparison between the three savers survives that intact, because they are all being deflated by the same amount.
• No fees, and no tax drag, which is roughly true inside a Roth and distinctly untrue in a taxable brokerage account.
• Contributions land at the end of each year, which is the conservative assumption. Paying in monthly through the year does slightly better.

None of that changes the shape of the result. A 1% fee, or a lower return, moves every one of these numbers, and moves them all in the same direction.

What to do with this

If you are in your twenties, the useful takeaway is not “save more”, which is not advice anybody is short of. It is that the account looking unimpressive after five years is not evidence of anything. That flat stretch at the left of the first chart is the part that ends up mattering most, and it is also the only part that feels like it isn’t working.

If you are past that and reading this with a certain amount of irritation: the same maths runs forwards from wherever you are now. The gap between starting today and starting in three years is smaller in dollars than the gap between 20 and 30, but it is the same shape, and it is still the largest single lever you have that does not require guessing what the market will do.

Either way, the version worth looking at is your own. Put your real age, balances and contributions into the calculator and open the Growth tab; it draws exactly the charts above, from your numbers instead of these.

It is, however, genuinely hard to answer. The money is spread across accounts with different tax treatments and different withdrawal rules. Social Security depends on when you claim it. Required Minimum Distributions start on a date that depends on the year you were born. A pension may or may not keep pace with inflation. And all of it has to be measured against a spending number that is itself only a guess.

This site is a calculator that holds all of that in one place and projects it forward, one year at a time, until you either run out of money or you don’t.

How to start

Open the calculator and work through the guided setup. It asks for your age, when you would like to retire, the accounts you hold and roughly what you expect to spend each year. That is enough to produce a projection.

Everything stays editable afterwards. The setup exists so that a first-time visitor is not confronted with eleven tabs and no idea which one to open; it is not a one-time gate, and re-running it is a reasonable way to change several assumptions at once.

What it models

• Year-by-year balances for pre-tax 401(k), 403(b), 457(b) and TSP, Roth, traditional IRA, HSA, taxable brokerage, pensions and rental property.
• Social Security at a claiming age you choose, indexed to inflation.
• Pensions, with an optional partial COLA, because most are not indexed, and pretending otherwise is the most common way a plan looks fine and isn’t.
• Required Minimum Distributions from your SECURE 2.0 applicable age, which is 73 or 75 depending on your birth year, covering pre-tax and governmental 457(b) balances.
• Taxes on withdrawals, grossed up. If you need $60,000 to spend, the plan withdraws enough that $60,000 survives the tax.
• Monte Carlo simulation over randomised annual returns, so you get a probability that the plan holds rather than a single deterministic number that implies more certainty than anyone has.
• Households where two people retire in different years, each with their own retirement date, RMD clock and claiming age.
• Named scenarios, “retire at 60” against “retire at 65”, compared side by side.

Every figure comes from one projection engine, so the Overview, the RMD tab and the Monte Carlo simulation are all reading the same numbers rather than each computing their own.

What it does not do

Being clear about this matters as much as the feature list:

• It is not advice. It is a model that does exactly what you tell it to. Change the assumed return rate and the answer changes, which is the point.
• It does not model tax brackets, only a single effective rate. That is fine for planning and no substitute for a tax preparer in the year you withdraw.
• It does not know about healthcare before Medicare, beyond whatever you put in your spending figure. The gap between retiring at 60 and turning 65 is real and expensive, and this tool will not remind you about it.
• It does not account for your specific circumstances: a divorce decree, a disability, a business sale, an inheritance. It models the ordinary case.

Your data stays in your browser

There is no signup, no account required, and no server doing the maths. The projection runs in your browser, and everything you enter is stored in that browser’s local storage. Nothing is transmitted anywhere.

That is also why the site needs JavaScript enabled, and why clearing your browser data clears your plan.

There is an optional paid tier that adds cloud sync of the same figures across devices, for people who want to pick up on a phone where they left off on a laptop. Every projection, scenario and Monte Carlo feature is free and always will be. The paid tier adds sync, not maths.

What this blog is for

Mostly, explaining the assumptions.

Any retirement calculator is a pile of assumptions wearing a single number as a disguise, and the useful thing is usually not the number but knowing which assumption it is resting on. So these posts work through them one at a time: what the projection actually assumes, how a household where two people retire in different years is modelled, and more as they come up.

If you would rather watch than read, the app has short narrated walkthroughs of each feature under its Videos tab, and a How It Works tab that explains every formula in plain language.

Most retirement calculators cannot describe that period at all. They ask for a retirement age, put every account in one pile, and switch the whole household from “saving” to “spending” on a single date.

Why one pile is wrong

The problem is not precision, it is direction. Collapsing the household into one portfolio makes two errors that both flatter the plan:

1. It lets the plan spend money it cannot reach. Your spouse’s 401(k) is not available to fund your spending at 62 without penalty. A single pooled balance quietly assumes it is.
2. It stops contributions too early. Your spouse is still working and still contributing for another seven years. Ending all contributions at the first retirement throws away seven years of deferrals and match.

Net of the two, the pooled model tends to look more achievable than the real thing in the early gap years and then run short later.

Accounts carry their own rules

The fix is to stop treating money as four buckets (pre-tax, Roth, taxable, HSA) and start treating it as pools that belong to someone. Each pool carries the rules that govern it: who may draw from it, until when it receives contributions, and which RMD clock it is on.

During the gap, then:

• The plan draws only from pools belonging to whoever has actually retired.
• It keeps contributing to the still-working partner’s accounts.
• Jointly held money is drawable from the first retirement but keeps receiving contributions until the last one.

That last rule is the one that looks inconsistent and is not. A joint brokerage account genuinely is spendable the day either of you stops working, and it genuinely does keep growing from the salary that is still arriving.

The safety property

There is an obvious risk in adding this machinery: that it changes the answer for the millions of households it should not apply to. A single person has no gap year, and their projection must come out exactly as it did before.

So pools whose rules match get merged before the projection runs. A single-owner household collapses to one pool and produces bit-for-bit the same numbers as the old four-bucket model. The staggered logic only ever engages when there is actually a stagger to model.

What to do with this

If you are modelling a couple with an age gap, the two inputs worth being deliberate about are the spouse’s retirement age and the spouse’s Social Security claim age. Both default to something reasonable, and neither default is right for everybody. The gap years are where a plan is most fragile and where the assumptions are least visible. That is a bad combination to leave on autopilot.

Today’s dollars, inflated forward

The two figures you enter for spending and Social Security, retireSpend and ssIncome, are today’s dollars. Internally they are inflated forward from your current age to the year being projected, so a $70,000 spending target at age 45 is not $70,000 at age 70; it is whatever $70,000 buys then.

This matters because the alternative convention, entering future dollars, is almost impossible to do consistently. Nobody has an intuition for what their grocery bill will be in 2051.

Social Security gets a full COLA. Pensions do not.

Social Security benefits are indexed to inflation, so the projection grows them at the full inflation rate every year. Most private pensions are not indexed at all, and public ones are usually partially indexed. So pension income is grown only by whatever COLA you give it, defaulting to none.

Over a thirty-year retirement that difference is enormous. A $30,000 pension with no COLA against 2.5% inflation is worth about $14,300 in today’s money by year thirty. Treating it as inflation-proof is the single most common way a plan looks fine on paper and does not hold.

Withdrawals are grossed up for tax

If the plan needs $60,000 of spending money out of a pre-tax 401(k), it does not withdraw $60,000. It withdraws enough that $60,000 survives the tax, at the effective rate you supply. Calculators that skip this step systematically overstate how long a portfolio lasts, because they quietly spend the government’s share too.

RMDs start at your applicable age, not at 70½ or 72

Required Minimum Distributions have moved twice in recent years, and a lot of material online is still out of date. Under SECURE 2.0 the applicable age depends on when you were born:

The projection derives this from your birth year rather than hardcoding a number, and applies it to pre-tax balances and governmental 457(b) balances, which are easy to forget.

Couples run on two clocks

If you are modelling a household, the projection is indexed to one person’s age, the primary’s, but the spouse has their own retirement age, their own RMD cohort and their own Social Security claim age. All of it gets converted onto the shared axis.

One consequence is worth stating outright, because the default surprises people: leaving the spouse’s retirement age blank means “retires in the same calendar year you do”, not “retires at the same age number”. For a couple six years apart those are wildly different plans, and the second one invents a six-year staggered window that nobody asked for.

What it does not model

Being explicit about the gaps matters as much as the assumptions:

• Sequence-of-returns risk is not in the deterministic projection at all. It assumes a steady annual return. Use the Monte Carlo tab for that.
• Tax brackets are a single effective rate, not a bracket schedule. Good enough for planning; not a substitute for a tax preparer in the year you actually withdraw.
• Healthcare before Medicare is only whatever you put in your spending figure. The gap between retiring at 60 and turning 65 is a real and expensive thing this tool will not remind you about.