· 5 min read
Real-transaction budgets beat forecasts — every time.
A forecast is a guess dressed up as a plan. A budget measured against your real transactions is just a fact — and that’s exactly what makes it useful.
Most budgeting apps ship with some version of a forecast. “You’ll spend €480 on groceries this month.” “You’re on track to overspend by €62.” “Your projected savings are €170.” It feels smart. It feels almost magical.
Most of the time, it isn’t.
The problem with forecasts
A forecast is a model stacked on top of your actual behaviour. The simpler version is a run-rate: take what you’ve already spent this month and extrapolate. The fancier version involves trends, regressions, seasonality — sometimes with an “AI” sticker slapped on.
Two things go wrong with forecasts, almost always:
- They mislead early in the period. A forecast run on day 3 of the month will double your annual pizza budget because of one Friday evening. That number has no predictive value — but it looks like one.
- They lose credibility as the data grows. By day 28, your forecast converges on the real total. At that point, you didn’t need a prediction; you needed the real number. Forecasts solve the wrong problem.
There’s a deeper issue too: a forecast is ambiguous. Is the number telling me what I will spend, or what the algorithm guesses I might spend if I act exactly like last month? Most interfaces don’t say.
What a real-transaction budget actually does
A budget in BudgetLuma is a fact. It says: “Here’s the amount I chose to spend on this wallet (or this category) over this period, and here’s the exact sum of transactions I’ve recorded so far in that period.”
No estimate. No model. Just two numbers:
- The target you set.
- The exact amount your wallet shows, in the same currency.
Both numbers are grounded. The target is something you decided. The progress is something that actually happened. There’s no interpretation layer; no black box; no “the algorithm says”.
Forecasts still have a place
To be fair, we didn’t delete them. BudgetLuma’s Insights screen can show a monthly run-rate projection — but only when it’s meaningful. There are strict hide conditions: if there isn’t enough history, if the month is too young, if there are fewer than five transactions in the period, the forecast stays hidden. An empty chart beats a misleading one.
When it does show up, it’s labelled as a projection, not a prediction, and it’s visually separate from your budget progress. They answer different questions.
Why this scales to shared money
A real-transaction budget is also the only model that works when multiple people are spending from the same wallet. If two partners split household expenses, a forecast has to guess how each person will behave; a real-transaction total just adds up what they’ve already done. Your partner putting groceries on the card at 18:02 shows up in the progress at 18:02, with no extra sync, no reconciliation step.
That’s also why we made BudgetLuma budgets work natively on shared wallets. It’s the same mechanism: budget attached to a wallet, progress equals the sum of the wallet’s transactions for the period. No special case. No fragile split logic.
The summary, one line
A budget isn’t a prediction. It’s a choice plus a thermometer. Keep the thermometer honest — by measuring the real temperature — and the budget stays honest too.