By Antonio Hill
Somewhere between 20 and 25 percent. Of one account. Not of your whole portfolio.
That’s the answer, and the reason it sounds strange is that every allocation chart you have ever seen got drawn for somebody who stops working at 65 and needs the money to last 30 years. You want to stop at 40 and make it last 50. Different problem. The standard advice doesn’t just miss by a little here. It points you the wrong way.
The Number
Search this question and you get three answers on repeat. Subtract your age from 100. Or from 110, if the site is feeling generous. Or a brokerage table telling you that from 60 to 69 you want 60 percent stocks and 35 percent bonds.
Run any of them at 40 and you land near 30 percent bonds.
Thirty percent. On a portfolio that has to survive five decades, two recessions nobody has invented yet, and every year you spend not earning a paycheck.
On this site everything comes back to three levers. Time in the market, what you put in, and what it grows at. In retirement those levers run backwards. Withdrawals replace contributions. Time in the market becomes the number of years your money has to survive. And sequence of returns risk is the return lever swinging at you at the worst possible moment. Every age-based rule of thumb you just read solves for that one bad moment by permanently taxing the return lever for fifty years. That trade makes sense when you have 30 years left. It’s a terrible trade when you have 50.
The Bond Tent Got Built In A Simulation
The strategy behind all of this has a name. It’s called a bond tent, and once you see it you can’t unsee it.
Picture your bond allocation on a chart across your life. It sits near zero while you’re working. Then in the ten years before you quit, you sell stocks and buy bonds until bonds are a big chunk of the pile. That’s the climb. Then in retirement you spend the bonds down and drift back toward stocks. That’s the descent. Draw it and you get a tent.
The logic is fine. Your portfolio is biggest right when you stop adding to it, so a crash at that exact moment costs you more real dollars than a crash at any other point in your life. Michael Kitces called that window the retirement red zone and wrote in 2014 that the optimal shape looks like “a V-shaped equity exposure.” He and Wade Pfau ran the numbers and landed on 30 percent equities at retirement, gliding up to 70 percent over the following 30 years.
Then Karsten Jeske at Early Retirement Now took that exact glide path and tested it against real market history instead of a simulation. It came out as one of the worst performers he ran. Not just worse than the other glide paths. Worse than most plain fixed allocations, meaning you would have done better picking one stock and bond mix on day one and never touching it again. His words, from SWR Series Part 20 in September 2017, are that the Kitces and Pfau path is “pretty universally inferior, even over a 30-year horizon.”
Read that again. It loses at 30 years. The horizon it was designed for.
So why does something that clean on paper fall apart on real data? Because Kitces and Pfau used Monte Carlo simulation, and Monte Carlo has no memory. It draws each year’s return out of a hat, so the odds after a 50 percent crash look identical to the odds after a 50 percent rally. Real markets have never behaved that way. After a deep drop you usually get a violent recovery. In the twelve months off the March 2009 bottom the S and P 500 returned more than 72 percent. A model that can’t see that coming will always tell you to keep too much in bonds for too long, because it never gets to show you the reward for climbing back in fast.
That’s the whole failure. The tent isn’t calibrated to the wrong retirement length. It’s calibrated to a market that doesn’t exist.
Where I Have To Check Myself
I’ve cited Guyton and Klinger on this site before. Their 2006 paper in the Journal of Financial Planning is the source of the number a lot of people in this space lean on, that you can start withdrawing 5.2 to 5.6 percent and hold up over 40 years, as long as you keep at least 65 percent in stocks and actually cut spending when the rules tell you to.
That paper is Monte Carlo too. Same method. Same blind spot.
I’m not going to attack one study for its methodology in one paragraph and quietly lean on another study with the identical methodology three paragraphs later. So here’s my updated position. Guyton and Klinger built something useful, and the mechanism, cutting spending in bad years instead of pretending nothing happened, is real and I will still run it. But the specific number that came out of the model deserves the same skepticism I just aimed at the bond tent. Kitces himself has since published an analysis showing a Great Depression era retiree getting pushed into real spending cuts approaching 45 percent under that exact rule, a number I went deeper on in the guardrails breakdown. When the guy who did more than anyone to popularize dynamic withdrawals shows a 45 percent cut coming out of his own framework, listen. Treat 5.2 percent as a ceiling somebody’s model produced, not as a floor.
The slow leaks nobody warns you about
A portfolio that’s too conservative for your timeline is one of ten quiet mistakes that add years to your working life without ever announcing themselves. I put all ten in a free guide, with the math on what each one actually costs.
The Shape That Actually Worked
Jeske didn’t stop at tearing the tent down. He tested 32 different glide paths across more than 1,700 historical retirement cohorts, and a clear winner showed up. Start around 60 percent equities at retirement. Climb to 100 percent. Get there in roughly ten years, not thirty.
Three things about that result.
The endpoint carries most of the weight. Whether you start at 40 percent or 60 percent or 80 percent barely moves the outcome. Whether you finish at 100 percent moves it a lot. If poor returns beat you up in your first fifteen years, going all in for the next fifteen is often the only thing that digs you out.
The speed matters almost as much. Ten years covers roughly one full bear market plus the bull that follows it. Dragging the same climb across thirty years means you’re still sitting in bonds long after the discount is gone.
And it’s conditional on valuations. When the Shiller CAPE ratio sits below 20, glide paths add nothing and something close to 100 percent stocks the entire time wins outright. Above 20, the glide path earns its keep. As of July 28, 2026 the CAPE is 40.57. That’s not a gray area.
One honest caveat, because I’d rather you hear it from me than find it later. The improvement is real and it’s modest. Over a 60 year horizon with an elevated CAPE, Jeske found the safe withdrawal rate moved from 3.25 percent to 3.47 percent. That’s roughly a five percent bump in what you get to spend. It does not rescue the 4 percent rule over fifty years and anyone telling you a glide path does that is selling something.
You Don’t Have One Portfolio. You Have Two.
Here’s where every study I just cited stops being enough, and where the whole thing clicked for me.
Kitces and Pfau model one portfolio over 30 years. Jeske models one portfolio over 60. Every result on page one of Google models one portfolio and an age. But if you retire at 40, you don’t have one portfolio. You have two piles of money doing two completely different jobs.
Pile one is your taxable bridge account, and it starts working the day you quit. It funds groceries. It gets sold into whatever the market hands you.
Pile two is your 401k and your Roth. You will not touch that money for the better part of twenty years while you work out how to reach it before 59 and a half. It has no near term job at all.
Sequence of returns risk only bites the pile you’re selling from. Which means the buffer belongs in the bridge, and the retirement accounts stay at 100 percent equities the entire time. Not most of it. All of it. Putting bonds in a 401k you won’t open for two decades is paying an insurance premium on a car you keep in the garage.
And that changes the number you were looking for, because the percentage depends entirely on what you divide by.
| Three years of spending in cash, measured against | Comes out as | Reads as |
|---|---|---|
| Total portfolio, bridge plus retirement accounts | 15% | Reckless |
| The bridge account you actually spend from | 25% | Textbook |
Identical dollars. One of those numbers gets you talked out of a good plan by somebody with a spreadsheet, and the other one lands right where Jeske puts it. Asked directly what allocation a person drawing down assets needs, his answer was about 20 to 25 percent in safer assets.
The rule. Size your buffer against the account you’re actually drawing from, never against your net worth. Three years of spending held against a bridge account usually lands between 20 and 25 percent, which is exactly where you want it. Held against your total portfolio it looks like nothing, and you’ll talk yourself into buying bonds you don’t need.
Here’s my own number, because a range is easy to hide behind. Three years of spending is about $360,000 for us. It sits in the bridge account and nowhere else, and I hold zero bonds. That’s a little under a fifth of the bridge, right where the research lands, and I got there before I’d read a word of the research.
Now the part I’m supposed to leave out. What I’d tell a reader in my exact seat is not identical to what I’m doing. The research supports a few percent in bonds alongside that cash. I’m holding none of it. That gap is not a better model, it’s a temperament. I’m greedy, my risk tolerance is higher than almost anyone’s I’ve met, and I want to make as much money as I possibly can. Those are reasons. They are not arguments. Take the cash buffer, and if a few percent in bonds is what lets you leave the rest alone through a bad decade, take that too.
Buckets Only Work If You Actually Drain Them
The bucket strategy gets sold as the friendly version of all this. Keep a cash bucket, a bond bucket, a stock bucket, spend the cash first, sleep at night.
Jeske took that apart in Part 48 and the critique is hard to argue with. If you refill the cash bucket every good year by selling stocks, you’re rebalancing. And rebalancing quietly pulls money out of the exact assets the buckets were supposed to protect. Money is fungible. Label a dollar however you want, it doesn’t care.
Buckets earn their reputation only when you let the cash bucket run down and stay down. Which, if you look at what’s actually happening to your allocation while that plays out, is a glide path, isn’t it?
The Ramp Builds Itself
Now put the two pieces together, because this is the part I like most.
Your buffer sits in the bridge. Your 401k and Roth sit at 100 percent equities. You spend the bridge down over your first years of retirement, and you leave the retirement accounts alone.
Watch what happens to your total equity allocation while that runs. The bridge shrinks. The retirement accounts keep compounding. Your equity percentage climbs on its own, from somewhere in the eighties toward 100, over roughly the window the research says it should.
You never rebalanced. You never set a calendar reminder for a glide path you’d have to remember to execute every month for a decade while a bear market screams at you to do the opposite. You spent from one account and ignored the other.
That’s the whole trick and it’s the same trick that works everywhere else in this game. I’ve never once beaten a market with willpower. I’ve beaten it with automation and habits. Build the structure so the right thing happens by default and your discipline stops mattering so much.
What The Buffer Costs You Now, And Why That Changed
Every warning you’ve read about holding cash in retirement got written in a different world. Jeske ran his cash cushion simulations in 2021, when short term inflation adjusted yields sat around negative one to negative two percent. Holding a buffer meant paying a fee for the privilege.
Check the tape now. As of July 2, 2026, real yields on inflation protected Treasuries stood at 1.99 percent at five years, 2.26 percent at ten, and 2.79 percent at thirty, straight off the Treasury’s own daily par real yield curve. Above inflation. Guaranteed by the Treasury.
Same strategy. Opposite price tag. The buffer used to bleed you and now it pays you roughly two percent over inflation to sit there and wait. That single change is why I’d argue the case for a modest buffer is stronger today than at any point in the last fifteen years, at the same moment the case for a deep slow bond tent got weaker.
It also hands you a decision tool nobody talks about. Take whatever your bridge account holds and price a twenty year ladder of inflation protected Treasuries against it. That number is your floor, the amount you could spend every year with zero market risk. Anything you plan to spend above that floor is you consciously choosing to take equity risk to fund it. Not a vague feeling about risk tolerance. A dollar figure. Run it once and you’ll know more about your own plan than most people ever find out.
Do this today. Open your accounts and write down two numbers. What sits in the accounts you’ll spend from before 59 and a half, and what sits in the ones you won’t touch. Multiply your annual spending by three. If you do not know what the first pile will be worth on your quit date, the early retirement calculator sizes it in its Bridge Years card. If that number lands between 20 and 25 percent of the first pile, you’re done, and every dollar in the second pile belongs in equities. If it lands far outside that band, you found your problem, and it isn’t your bond allocation.
What This Actually Means For You
The advice you found is wrong, and it’s wrong in the expensive direction. Thirty percent bonds at 40 is a permanent tax on the one lever that has to carry fifty years of compounding, paid to insure against a window that lasts a few years.
You want a buffer. You want it small, roughly three years of spending. You want it in the account you’re spending from and nowhere else. And you want it gone in under a decade, with everything else at 100 percent equities the whole time, which happens by itself if you just leave the retirement accounts alone.
Somebody is going to read this and decide I’ve gone soft because I’m close to the exit. I don’t know how you’d get there. I’m as aggressive as I have ever been and I plan to stay the most aggressive investor most people will ever meet. The only thing that changed is that I’ll hold a bucket of cash when I quit, and that’s common sense, not a flinch. If you want to run 100 percent equities the whole way through with nothing set aside, good luck. You can absolutely do it. I’d just rather not depend on getting lucky.
The US market has recovered from every drawdown it’s ever had so far. That’s history, not a promise, and I’d never tell you to bet the plan on it repeating on schedule. But if you’re going to build a fifty year retirement around the assumption that stocks eventually come back, then flinching for thirty years in bonds is a strange way to show it. Take the small insurance policy. Then get back in the game and stay there.
Get the 10 Quiet Mistakes That Kill Your Early Retirement
A portfolio built for the wrong retirement is one of them. So are eight others most high earners never notice until the timeline has already slipped. Free, no fluff, real numbers on what each one costs. Send me the guide
Questions People Ask About This
What is a bond tent?
A bond tent means building up a large bond position in the years right before retirement, then spending it down afterward. Charted across your life, the bond allocation climbs, peaks at retirement, and falls, which looks like a tent. The idea protects you at the moment your portfolio is largest and most exposed. The problem is the version most people follow came out of a simulation that missed how markets actually recover, and it lost to plain fixed allocations when tested against real history.
How much cash should I have when I retire early?
Roughly three years of spending, held in the taxable account you’ll draw from before 59 and a half. For most early retirees that lands somewhere between 20 and 25 percent of that account. Be honest that three years does not cover the worst stretches on record. Jeske found the biggest benefit from cash buffers running six years in the 1929 case and eleven to thirteen in the 1973 case, and the mid 1960s cohort needed even more. A three year buffer buys you a 2020, not a 1966. You cover the rest with spending flexibility, or by working one more year.
Should I hold bonds in my 401k or in my brokerage account?
If you’re retiring early, hold them in the brokerage account you’re actually spending from. Your 401k and Roth have no job for close to twenty years, so there’s nothing there for bonds to protect. Standard asset location advice says the opposite, put bonds in tax deferred accounts because interest gets taxed anyway, and that advice assumes you retire at 65 and touch everything at once. It doesn’t survive contact with an early retirement.
Does the 100 minus your age rule work if I retire at 40?
No. It puts you near 30 percent bonds when you have fifty years of compounding left and one brief window of real danger. Every version of that rule, whether it subtracts from 100 or 110 or 120, assumes retirement starts around 65 and runs about 30 years. Change the horizon and the answer changes with it. Read more on why sequence risk works differently over fifty years and what over diversifying costs you.
Disclosure. Structured Wealth is written by one person documenting his own path to early retirement, and links in this article point to research I actually use. The free guide and the paid Blueprint are my own products. I’m a mechanical engineer, not a licensed financial advisor, and nothing here is personalized investment advice. Every figure in this piece is sourced and dated. Verify anything before you act on it.
Return and yield figures current as of July 28, 2026. Withdrawal rate and glide path findings from Early Retirement Now SWR Series Parts 20 and 48. CAPE from multpl.com. Real yields pulled directly from the US Treasury daily par real yield curve (home.treasury.gov), dated July 2, 2026. Data referenced: Guyton and Klinger (2006), FRED 10 year TIPS series.
