The Onside methodology

We don't hand-pick the stars and guess at everyone else. The exact same engine values a Champions League forward and a third-choice goalkeeper in the second tier. That's the point: a level playing field, applied without favour, so two players with identical seasons get identical valuations regardless of how famous they are.

The model rests on a handful of pillars. Each one captures something a scout would actually care about, and each is weighted differently depending on the position — output matters far more for a striker than for a centre-half, and we treat them accordingly.

A point estimate on its own is a kind of lie — it pretends to a precision nobody has. So we never show one alone. Every valuation ships with a band: a low and a high we genuinely believe the player sits between.

The band's width is honest about what we know. When we hold a full, clean season — minutes, ratings, goals, age — the band is tight and confidence is high. When a player is barely featured, has no reliable rating, or is missing an age, we widen the band and lower the confidence rather than fake a number. Less data, more humility. You always see exactly how sure we are.

Football doesn't stand still, so neither do our valuations. Each day the board ticks via the Pulse — a small, deterministic daily movement layered on top of the underlying model. It's what makes the market feel alive: a player you watched yesterday will read a little higher or lower today.

Deterministic is the operative word. The Pulse isn't random noise and it isn't a trader on a desk nudging numbers. Given the same inputs, it produces the same movement, every time — fully reproducible, auditable, and free of human thumbs on the scale.

The reference point most fans know — Transfermarkt — builds its values from community debate: users propose figures, moderators arbitrate, and a number eventually settles. It's a remarkable feat of collective knowledge, and we have real respect for it. But it is, by construction, an opinion poll. It inherits the crowd's biases, moves on sentiment, lags reality, and can't tell you why a number is what it is.

Onside takes the opposite approach. The same model runs on every player, the inputs are stated, the pillars are visible, and the maths is the maths. You can disagree with our weighting — but you can see it, and you'll always get the same answer from the same data. That's the trade we're making: transparency and consistency over the wisdom of the crowd.

A model is only as good as what it can see, and we'd rather name the limits than pretend they don't exist:

• ●Advanced metrics aren't everywhere yet. We don't have live expected-goals (xG) and shot-quality data for every league we cover. Where it's missing, the model leans harder on the signals it does have, and the confidence band reflects that.
• ●National-team valuations are squad-based. A country's value is built up from the players in its pool, not from a separate international model. It's a useful lens for a tournament, but it's an aggregate, and we label it as one.
• ●We don't model every intangible. Injuries, release clauses, a falling-out with a manager, off-pitch noise — the things a human scout weighs that don't show up cleanly in a stat line. The Pulse and the bands soften this, but they don't erase it.

We'd rather lose an argument honestly than win one by hiding the method. As coverage deepens — more leagues, richer data — the bands tighten and the estimates sharpen. The model improves in public, on the record.