How Accurate Is a Financial Forecast, Really?
"Nobody can predict the future" is a fair objection to a financial forecast, and also slightly beside the point. This is what actually drives the accuracy of a household forecast, and where the real uncertainty lives.
The most common objection to any financial forecast is a reasonable one: nobody can predict the future, so what's the point? It's worth separating out what a household forecast is actually trying to do, because most of it isn't prediction at all.
Two very different kinds of number
A forecast is really two categories of number bolted together. The first is mechanical: your income tax and National Insurance given a salary, the interest and capital split on a repayment mortgage, the remaining balance on a student loan, a pension contribution as a percentage of pay. These aren't predictions. They're calculations, and given the right inputs they're either right or they're a bug. The second category is genuinely uncertain: investment returns, house price growth, whether you get that promotion. No model, CrestCast included, can predict those with certainty.
Where CrestCast focuses
CrestCast is built to get the mechanical category exactly right: income tax on either the rest-of-UK bands or the six Scottish ones, whichever applies to you, National Insurance, all five student loan plans, pension contributions, and your specific loan repayment schedules. Given accurate inputs, the monthly figures should match reality closely, because that part genuinely is just arithmetic done correctly, applied to your actual numbers.
The variable that actually matters most
The biggest source of inaccuracy in any forecast isn't the model, it's stale inputs. A forecast built on last year's salary, an old bill amount, or a mortgage rate that changed six months ago will drift from reality, not because the maths is wrong but because the world it's describing has moved on. The single highest-leverage thing you can do for accuracy is update the model when something actually changes.
A forecast you revisit, not a one-off prediction
That's also why CrestCast lets you save your plan as a named version and compare it against a later one. Instead of treating a forecast as a single prophecy that's either right or wrong, you can see how reality tracked against the plan, adjust the inputs, and keep the model current.
›About this post
This post is about the nature of forecast accuracy itself (the split between mechanical calculation and genuine uncertainty), which doesn't have a natural chart or screenshot to attach. A picture of a forecast doesn't illustrate the argument any better than the explanation above already does.
To see the mechanical side in action, with tax, National Insurance and mortgage amortisation calculated against real numbers, see "The Three-Way Financial Model, Explained" or any of the scenario-comparison posts on this blog.
See how your own forecast tracks reality
CrestCast forecasts your household from your own figures, and pinning what actually happened keeps the model anchored to reality rather than drifting away from it.
Try CrestCast free →Interested in CrestCast?
Create your free account →