Market Values

Market values are the price tags everything financial hangs off: affordability in the recommendation shortlist, the targets the valuation source models learn, the search filters, and the figure on every player page. Oraca does not guess a value — it uses the observed market consensus, stamps one canonical number on each player, and keeps a clear record of where that number came from.

In one line: every screen reads one value per player, and that value is written by a single process that resolves primary community valuation first, a secondary feed where the primary doesn't reach, and the last-known figure as a safety net. The underlying source histories are never overwritten, so the canonical value is always reconstructable and its provenance is always known.

Why observed, not estimated

You can either predict a player's value from his attributes or observe what the market actually says it is. For the headline figure we deliberately observe — a well-studied, crowd-sourced market consensus is more trustworthy as a present-day price than any attribute model, and it doesn't smuggle our own assumptions into the number. Forecasting where that value is heading is a separate job, handled by the dedicated valuation models — and even there, we withhold any euro figure that isn't yet calibrated well enough to trust.

One value, one writer

Dozens of surfaces need a player's price, so it's stored once per player rather than re-derived on every query. The risk with a stored value is drift — many writers, inconsistent results — so exactly one process owns the column. It resolves, in precedence order:

1. the latest high-confidence primary valuation, where we can confidently match the player;
2. failing that, the latest figure from a secondary feed, which fills players the primary source doesn't cover;
3. failing both, the player's last-known value, so he never silently drops to nothing.

The two underlying source histories are retained and never mutated — the writer only reads them — so the canonical value is fully reproducible by re-running the resolution, and you can always see which source and which date produced it.

Precision-first matching

A valuation is only useful if it lands on the right player. Matching a player to an external valuation source is itself an identity problem (see Identity Resolution), so we take only high-confidence matches — typically an exact date-of-birth agreement — and quarantine anything weaker. A player whose match isn't high-confidence simply falls through to the secondary feed or his last-known value, rather than risking a wrong price on the wrong man. The depth of history this gives us — valuations running back roughly two decades for well-covered players — is what feeds the forward-looking valuation models.

How we keep it honest

This is provenance, not a model, so honesty here is about source trust and freshness:

• Source trust — the primary source is an externally studied market consensus, and only high-confidence matches are used, so a trusted figure lands on the right player.
• Reconstructable by design — because one process owns the value and never mutates the source histories, the canonical number can be rebuilt from scratch at any time, and its origin is always inspectable.
• Observed, not predicted — the headline value is the market's price today; forecasting where it goes is a separate, clearly-labelled job.

What it can't do

• Current-value freshness varies by source. The secondary feed refreshes on a fixed weekly cadence; the primary source is refreshed as players are re-checked, so a given player's headline figure is as fresh as his last refresh — the history is deep, the latest point can lag.
• High-confidence-only matching leaves gaps. A player whose identity doesn't match cleanly is valued from the secondary feed or his last-known figure rather than the primary source.
• It's a price, not a fee. Affordability uses market value as a proxy for a transfer fee; an actual fee can diverge sharply from it.
• Coverage, not model. A player in a competition we don't cover has no valuation to read.

The research behind it

The design rests on the finding that crowd-sourced football valuations are reliable enough to use as ground truth, rather than estimating value from attributes.

• Herm, S., Callsen-Bracker, H.-M. & Kreis, H. (2014). When the crowd evaluates soccer players' market values: Accuracy and evaluation attributes of an online community. Sport Management Review. — Validates that community valuations are accurate, justifying their use as the observed value.
• Müller, O., Simons, A. & Weinmann, M. (2017). Beyond crowd judgments: Data-driven estimation of market value in association football. European Journal of Operational Research. — The data-driven estimation alternative; we use observed values for the canonical figure and reserve modelling for the forward-looking valuation models.