After I put the Roth conversion tool online, the question I kept getting wasn’t about conversions at all. It was simpler and scarier: “OK, but how much can I actually spend?” Almost every withdrawal rate calculator answers that by multiplying your savings by 4% and calling it a day. So I added a mode to my own optimizer that throws the flat percentage out and solves for the real number — the most you can spend every year while drawing every account down to zero by the end of the plan.
This builds on my 4-part series about optimizing IRA-to-Roth conversions in early retirement:
The trouble with a flat 4%
The 4% rule is a fine napkin estimate, and I’m not here to dunk on it. But it’s a single number applied to a single pile of money, and almost nobody actually retires with a single pile of money. You retire with a traditional IRA the government will tax on the way out, maybe a Roth that’s tax-free, and a brokerage account with embedded gains. A flat percentage treats all three as the same dollar. They are not the same dollar.
It also ignores timing. Social Security shows up at some age and changes everything. Required minimum distributions land at 73 or 75 whether you want the money or not. Other income stops when you stop working. A flat withdrawal rate can’t see any of that — it just draws the same slice every year and hopes the tax bill behaves. (It rarely does.)
What the “spend to zero” mode actually does
I’ll be honest about the detour I took to get here. I originally built the tool to do the responsible-sounding thing: maximize how much wealth you leave behind after taxes. It worked. And then I looked at my own output and realized I’d answered a question almost no early retiree is actually asking. Nobody emails me asking how to die rich. They ask how much they can safely spend without running out. So I added a checkbox that flips the whole problem on its head.
When you tick “find the most I can spend,” the model stops trying to preserve money and instead treats your annual spending as the thing to maximize. It finds the highest constant, inflation-adjusted amount you can pull every single year such that all three accounts land at roughly zero on your final year. Under the hood it’s the same linear program from part 3, just pointed at a different goal: instead of “maximize what’s left,” it’s “maximize what you spend, subject to ending with nothing.”
The part that surprised me is that the conversions don’t go away in this mode. You’d think drawing everything to zero would make Roth conversions pointless, but the optimizer still does them — because shifting income into your low-bracket early years shrinks your lifetime tax bill, and every dollar you don’t hand the IRS is a dollar you get to spend instead.
What it told me with real numbers
I ran a case close to a textbook early-retiree setup: single, age 62, planning to 90, with $1.2M in a traditional IRA, $50k in a Roth, and a $100k brokerage account carrying about $50k of embedded gains. Social Security of $15k starting at 70, some other income until then, a 5% real return, and planned spending of $60k a year.
In the normal “leave the most behind” mode, that plan funds the $60k and still finishes with roughly $670k sitting in the accounts at age 90. Which is great if your goal is an inheritance — and a little maddening if your goal was to actually enjoy the money. So I flipped on the spend-to-zero mode, and the number came back at about $70,000 a year — roughly ten grand a year more than I’d penciled in, every year for nearly three decades, with the accounts gliding down to zero right on schedule. (Seeing that gap between “what I planned to spend” and “what I could spend” was a genuinely uncomfortable moment.)
That’s the whole point of doing this with an optimizer instead of a percentage. Four percent of this portfolio would have spat out a confident, tidy, and badly wrong answer, because it would have taxed the IRA dollars the same as the Roth dollars and ignored the conversion savings entirely.
Why this beats a percentage every time
The reason a real withdrawal rate calculator should outperform a rule of thumb comes down to three things the rule can’t handle: which account a dollar comes from, what tax bracket you’re standing in when you pull it, and what forced income (Social Security, RMDs) is about to stack on top. The model decides the withdrawal order for you — spending the taxable account first, leaning on the IRA while you’re in the low brackets, and saving the tax-free Roth for last unless tapping it earlier genuinely lowers the lifetime bill. A spreadsheet full of 4% can’t reason about any of that; it doesn’t even know the IRS exists.
None of this is a promise, to be clear. It’s a model, and a model is only as honest as the assumptions you feed it — your return won’t be a smooth 5%, tax law will change, and your real spending will be lumpy in ways no constant number captures. But as a starting point it’s a far better one than a flat percentage, and unlike the spreadsheet it came from, you can run it yourself in a browser.
The spend-to-zero option is rolling out as a checkbox right under the spending input in the Roth conversion and drawdown tool — tick it, leave the spending box alone, and it’ll hand back the biggest sustainable number for your accounts. If there’s an input you wish it handled (ACA subsidies, IRMAA, state tax), tell me in the comments and I’ll try to fold it into the next version.
One disclaimer, because this is money and taxes: I am an operations research analyst, not a CPA or a financial advisor, and nothing here is tax or investment advice. This is just how I think about my own situation. Tax rules change and your situation is different from mine, so check the primary sources and talk to a professional before you actually move money.
Things that I use, like, and am affiliated with:
Mint Mobile offers great cell phone service for $15 flat, get $15 off using the link. Get discounted phones with service activation and no contract.
I never spend money before I check Mr Rebates or Rakuten to get cashbacks, rebates, discounts, coupons or cheaper gift cards.