The editorial process: artificial intelligence and human supervision
GoalCast AI content is generated with the help of artificial-intelligence systems from official sports data, and supervised by the editorial team that designs its rules, limits and quality checks. No text is written by hand match by match: the editorial team works on the system, not on the single article.
This applies to AI Analyses, match reports, news in brief, player profiles and market-value estimates. Each family of content has its own rules, documented on the pages linked at the bottom.
Results, statistics, line-ups, standings and odds come from professional sports-data providers and are updated several times a day. News in brief are summaries of articles published by sports outlets: they always cite the original source and add no facts that are not in the source.
Market-value estimates are computed by our model and verified against real transfer fees: the methodology page explains how, and the public verification shows where the model gets it wrong.
The AI processes the data and produces the texts according to explicit rules: it cannot cite regulations from memory, it cannot invent absences or line-ups, it cannot use words that promise a result (no prediction is "safe" or "guaranteed"). The editorial team defines the formats and the rules, checks the systems, spot-reads published content and acts on errors reported by readers.
No content is backdated or changed after publication without trace: publication and update dates are the real ones, and corrections follow the policy described on the dedicated page.
Every prediction is automatically compared with the real result at full time; every market estimate with real transfer fees. The outcomes stay public, errors included, on the accuracy page, with no selection of favourable cases.
If you find an error in a piece of content, report it to us from the contacts page: we check it and, if confirmed, we correct the content and, when needed, the rule that produced it.