Every ranking site claims to be accurate. Almost none of them ever check. The honest question is not whether a ranking sounds confident — it is whether it gets tested against reality, and how often it turns out to be right. Here's how to judge that, and how we hold ourselves to it.
A ranking that is never checked against reality is just an opinion in a nice font. The real test is prediction: turn the ratings into a forecast for a real game, then see whether it was right. A site that never makes a falsifiable call can never be wrong — and never proven right either.
The only fair test locks the forecast in before the whistle and grades it against the actual result, with no hindsight and no quiet edits afterward. Anyone can explain a result once they know it. Calling it in advance is the hard part — and the only part that counts.
It is not enough to be accurate in a vacuum; a model has to beat the alternatives. We back-test ours head to head against the standard approaches — a generic Elo rating, the academic Dixon-Coles scoreline model, a modern self-fitting regression, and simply sorting teams by their published national rank — on the same real, decisive games. In the states where we have validated it, ours calls more winners than the rest.
Even the best model deals in probability, not certainty. Youth soccer is volatile: a single bounce, a missing starter, or a hot goalkeeper can flip a result that the numbers got right on paper. An honest model is built to own that uncertainty rather than pretend it does not exist — a prediction is a well-informed lean, not a promise.
Rather than ask you to take our word for it, we publish real calls next to what actually happened, grouped week by week. See real predictions versus results.