What YNAB gets right — and the one thing it structurally can't model

If you keep long-dated money in a brokerage and have tried to keep YNAB honest about it, you have probably built some version of this: the brokerage brought on-budget as an unlinked cash account, so moving money into it doesn’t read as spending; a large Investing or fluctuation category held against the volatile part of the balance, so a bad month doesn’t quietly drain the categories behind it; a monthly pass that works out how much cash sits above your target and moves the excess in; and a reconcile whose adjustment you route into that buffer rather than letting it look like income.

It works. People run it for years — long enough to buy a car in cash without having to liquidate at a bad moment — and it is a genuinely clever piece of engineering on top of a tool that was not built for it. It is also four manual jobs a month, and the month you stop doing them the numbers stop meaning anything.

I want to be careful here, because the obvious next move is to say the tool is bad, and it isn’t. Zero-based budgeting is the thing that works. Give every unit of money a job, watch what is left, stop pretending the money in your account is available when it is already promised elsewhere — that insight is why any of us budget at all, and YNAB has taught it to more people than anyone. What follows is not an argument against it. It’s an argument about one axis it doesn’t have.

What the workaround is actually replacing

Look at each piece of that arrangement and ask what it stands in for.

The brokerage forced on-budget as a cash account stands in for the fact that an account has a job. It’s not an arbitrary container; it holds a particular kind of money, chosen for when that money is needed. YNAB has no way to say so, so the only lever available is the blunt one — on-budget or tracking — and neither is right. Tracking removes the money from the plan entirely. On-budget keeps it in the plan and makes every category behind it move with the market.

The buffer category stands in for the gap between what you have promised and what you actually hold. Your categories say you have twelve thousand set aside for the roof. The account holding it was worth twelve thousand in March and is worth eleven and a half now. Both numbers are true; the difference is a real fact about your position, and it has nowhere to live. So it gets parked in a category and called a buffer.

The monthly sweep stands in for the correction that closes that gap. And the reconcile adjustment you route by hand stands in for market movement being a different kind of event from income — because it is. Nobody paid you. The number moved.

Four chores, and they are all downstream of one missing column. YNAB knows what each amount is for. It does not know when you need it. And since when you need it is what decides where it ought to sit, the entire question of which account holds what falls outside the model — leaving you to hold it in your head, in a category, and in a calendar reminder.

The second axis

Duration’s claim is narrow: make the horizon a real axis and those three chores collapse into properties of the model. What follows is the shape of the idea rather than a set of instructions — if you want the instructions, moving from YNAB walks through the actual move, account by account.

• The on-budget invested account is an account with a horizon — a statement about when its money is needed, which is also a statement about what it may be held in.
• The buffer category becomes drift: the difference between what you have allocated and what your accounts are funded with, computed continuously rather than maintained by hand.
• The monthly cash-above-target pass becomes the sweep — the transfer that closes the drift, proposed when it’s worth doing.
• The reconcile adjustment becomes market movement, recorded as what it is rather than disguised as income and immediately corrected.

There’s a fifth thing that falls out of it, and for some people it’s the whole reason to look. If buckets are not tied to accounts, one invested account can back several of them at once — a new car, a roof, a trip in three years — each with its own balance and its own target, all inside a single holding you never have to split. That question, how do I see what amounts inside this account are assigned to what, is the one I have watched go unanswered in forum thread after forum thread. It has no answer in a model where an account is either in the budget or out of it.

What this is not

It is not an argument for investing money you are about to spend.

This needs saying plainly, because the moment horizons and investments appear in the same sentence, a reasonable reader hears put your emergency fund in an index fund — and they’re right to push back on that. The horizon axis is precisely the thing that says a repair fund due in March belongs in cash. That’s the axis working, not a caveat bolted onto it. Money you need soon has a short horizon; a short horizon means an instrument you can spend without caring what the market did this week. The whole apparatus exists to make that judgement explicit and visible, rather than something you re-derive every time you look at a balance.

It’s also not an optimiser. Duration holds no view on markets, recommends no allocation, and moves nothing on your behalf. It records what you decided and shows you the consequences, including the uncomfortable ones. If you want a tool that tells you what to buy, this isn’t it — and if you want a tool that tells you where you actually stand, that’s the entire design.

Who this is for, honestly

If YNAB serves you well and your savings sit in a savings account, you don’t have this problem and you should keep the tool you have. The monthly rhythm, the category view, the fresh start each month — those are real features and Duration deliberately does some of them differently, which suits some people far less than others.

But if you recognised your own spreadsheet in the first paragraph — if you have an Investing buffer category and a recurring monthly reminder and a mental note about which account is really holding the roof money — then you have already built this. You built it because the model was one column short, and you were right about which column.

Duration is that arrangement made native. It’s in invite-only use while we build it, and we’d rather hear from ten people who’ve run the by-hand version for years than a thousand who haven’t. If you want to see exactly what moving across would involve — what carries over, what you rebuild, and where the numbers deliberately disagree with YNAB’s — that is written up in moving from YNAB.