Why Steady Stops Forecasting at a Year

Kia ora. A reasonable question we get: PocketSmith will project thirty years, so why does Steady stop at twelve months? Isn't more better?

No, and it's worth explaining rather than hand-waving, because the reasoning is the product.

What a forecast is made of

A twelve-month projection in Steady is built from things that already happened to you. Your pay, its rhythm, your rent, your power bill's seasonal shape, the insurance that renewed in March, the rates instalments your council charges quarterly. Up to two years of your own transactions, which is what Akahu now provides.

Extend that to ten years and the ingredients change. There is no data about 2036. So the line has to be built from assumptions instead: your income grows at some rate, inflation runs at some rate, you don't change jobs or have a child or move cities.

That's not a longer forecast. It's a different kind of object wearing the same clothes.

Assumptions multiply

One assumption with a 10 percent error is fine. Six assumptions compounding over ten years is not — errors multiply rather than add, and by year seven the range is wider than the number is useful.

The uncomfortable part is that the chart doesn't show this. A thirty-year projection renders as a single confident line, at the same visual weight as tomorrow's balance. It looks more authoritative the further out it goes, which is precisely backwards.

Why exactly a year

A year is one complete seasonal cycle.

Every annual cost you have lands exactly once inside twelve months. All four rates instalments. One Christmas. One winter power season. One car rego, one WOF, one insurance renewal, one back-to-school February.

Stop at nine months and you'll miss things depending on where you start. Run eighteen and you've double-counted a Christmas you have no second data point for.

Twelve months is the shortest window containing everything, and the longest one you can build from evidence instead of guesses. Those two happen to meet at the same place, which is why it's the natural edge rather than an arbitrary one.

Being clear about the trade

We're giving something up. Long-range projection is genuinely useful for a specific set of decisions:

• Whether to salary-sacrifice more into KiwiSaver
• What a mortgage looks like over its full term
• Whether early retirement maths works
• Comparing two big financial paths over decades

Steady doesn't do those, and if that's your question, PocketSmith's multi-decade projections are the better tool. We'd rather say that than pretend.

What we're claiming is narrower: for the decisions people actually make weekly — can I afford this, will I make it to payday, what's coming that I've forgotten — a year built from real data beats thirty years built from assumptions. Those decisions are where the money is won or lost for most households.

What we do instead of going longer

Rather than extending the line, the effort has gone into making the twelve months honest:

Confidence that falls with distance. Every window states how confident it is, and it drops the further out you look. Day 340 is not presented with the same certainty as tomorrow.

Seasonality from your own history. Winter power, December, February. Learned from your transactions rather than a generic curve.

Annual bills on the timeline. Detected from history, including the once-a-year ones, which is the hard case — a yearly charge looks like a one-off until it happens twice.

Views at 7 days, 30 days, 6 months and 1 year. Different questions need different windows, and the short ones are the accurate ones.

The rule underneath

If we can't build it from your data, we won't draw it as though we can.

That costs us a feature comparison — "up to 12 months" reads smaller than "up to 30 years" on a table. We think a forecast you can trust for a year is worth more than one you can't trust for thirty, and that a chart admitting what it doesn't know is a feature rather than a shortcoming.

If you need the decades, use the tool built for decades. If you need to know whether Thursday works, we're better at that.