Missing is not zero: a field guide to incomplete football history

Historical football data rarely arrives as one complete career. Providers return rows by player, team, season and competition, and coverage can vary across those dimensions. When a response lacks an older season, the absence does not mean the player made zero appearances. Writing zero converts uncertainty about coverage into a false sporting claim and can corrupt totals, rates and readiness decisions downstream.

A zero is valid only when the source positively reports the measure as zero inside a covered row. Even then, the row needs context. FootIndex’s audit found competition entries with zero statistics beside populated rows for the same player and season. Such placeholders can describe registration, a provider catalogue match or a competition in which no recorded appearance occurred. They should not automatically become evidence of participation.

Names introduce another risk. The same tournament may appear under abbreviations, sponsor names or translated labels, while two genuinely different tournaments may have similar names. Normalising punctuation and case helps find possible duplicates, but merging on text alone can combine distinct competitions. Stable provider identifiers, governing body, country, season format and team association are safer reconciliation keys.

Friendlies require their own category. They are real matches and can be useful for workload or squad analysis, but combining them with competitive league and Champions League records changes the meaning of career output. The audited dataset contains international friendlies, club friendlies and a Club World Cup row whose recorded figures overlap with a friendly-labelled row. That is a signal for review, not an invitation to count both.

Transfers and loans make inference especially dangerous. A player can have rows for more than one club in a season, and a competition winner can be associated with only one of them. The same principle applies to appearances: a club name on a current profile cannot safely fill the team field of an old record. Historical association must come from the relevant dated row or a verified source.

FootIndex handles these states conservatively. Imported records retain source identifiers and timestamps; candidate duplicates and conflicting rows are quarantined for reconciliation; absent measures remain null; provisional profiles stay noindex and out of sitemaps. Publication requires enough reconciled season coverage to support the visible totals. This may leave fewer public pages, but prevents search engines and visitors from receiving invented completeness.

External competition and player pages are checkpoints rather than universal replacements. UEFA, the Premier League and LALIGA can help confirm scope and reveal a mismatch, yet their live pages may update after a frozen import. Every correction should record what changed, why, which source supported it and when the audit occurred. Transparency makes future automation safer because a revised value remains traceable instead of merely appearing different.