Behind Our Questions
Quiz sites usually write their questions by hand or have an AI write them. Most Goalasso games do neither: each round is assembled by code from football data, and your answer is checked against that same data. Here is where each mode gets its content, and where the limits are.
Three sources of content
- The player database: 9,318 players and 1,037 clubs, with nationalities, positions, birth dates and 23,902 club spells, compiled mostly from Transfermarkt data, with about 1,400 players added from Wikidata. Stadium Roar, The Bomb, Rank It, Tier List and Stay Quiet query it each time you play.
- Data packs built ahead of time: ValueGuessr (3,129 players in the big five leagues and the Saudi Pro League, with the Transfermarkt market values published by June 2026, most of them in late May and early June), TransferChain (2,707 players linked through big-five-league squads) and Career Path (thousands of careers read from English Wikipedia's footballer infoboxes, with the club crests stored on our own server) load files that we build with scripts and refresh by hand.
- Lists we wrote ourselves: Trophy Hunt's lists of past winners, captains, managers and referees, each checked against published records; Stay Quiet's fame scales for clubs and stadiums; and Tier List's pedigree tiers for clubs and nations.
Box to Box and Chain Reaction aren't quizzes: one is a 3D runner, the other a puzzle built on players' nationalities. The VAR is fiction, covered below.
Mode by mode
| Mode | Where the content comes from | How your answer is checked |
|---|---|---|
| Trophy Hunt | A hand-written list of ten past winners, captains, managers or referees, such as the 1990s Ballon d'Or winners, checked against published records | Your names are matched against the list by full name, surname or a known nickname, ignoring accents and capitals; a first name alone does not count |
| Stadium Roar | Two players and a stat computed from the database | Your pick is compared with the stored figures |
| The Bomb | A criterion such as a nationality, a position or a club | Each name is looked up in the database against the criterion; a partial name counts if it matches a player who fits |
| Rank It | Ten players and a criterion from the database | Exact positions against the order of the stored values, with an AI check on popularity rankings |
| Tier List | Players ranked by our popularity score; clubs and nations from our pedigree tiers | Points for how close you are to the expected tier |
| Stay Quiet | Players by popularity score; clubs and stadiums from our fame scales | Who counts as more famous is decided by those scores |
| Career Path | A player's senior career, read ahead of time from his English Wikipedia infobox, with each club's crest from its Wikipedia page | Your answer is compared with the player's full name, in English or French, or with a surname no other player in the game's list has, ignoring accents and capitals |
| ValueGuessr | The ValueGuessr data pack | Points for how close you are to the stored market value |
| TransferChain | The teammate links in the TransferChain data pack | The search only offers real teammates of your last player |
| The VAR | Incident cards written ahead of time | No right answer: your choices move the story's meters |
Where AI is used, and where it isn't
No AI model writes questions while you play. AI appears in two places only:
- Popularity rankings in Rank It. Popularity is our own estimate, so when a round ranks players by it, an AI model (gpt-oss-120b, run on Ollama Cloud) reviews the order and can correct it. If the model doesn't answer, the database order stands.
- The VAR's incident cards. 117 of the cards were drafted in English and French with the same model in September 2026 and checked by script for missing text and real names; we wrote the rest. The game engine, not the model, decides what each choice does, and every name in the game is invented.
Earlier Goalasso modes had an AI write their questions. They have been retired and can no longer be played.
Why we build questions this way
An AI writing questions can state something plausible that isn't true. Building rounds from data means every answer can be traced back to a record, and correcting one record corrects every round that uses it. The trade-off is that a mistake in the data becomes a mistake in the game, which is why the limits below matter.
Avoiding repeats
- Every mode avoids repeating a question within a game.
- Career Path remembers the last 50 players shown on your device and skips them while the level has others left.
- Between games, most modes draw at random, so the same player can come back, especially on Easy where the pool is smaller. Trophy Hunt's lists are fixed: the same topic always gives the same ten names.
Two languages
Questions use fixed sentence templates written in English and French. Player names stay as they are, and clubs and countries use their French names where we have them. Nothing is machine-translated while you play, and Box to Box is in English only.
Known limits
- Career histories are thinner for older careers than for recent seasons.
- Market values are snapshots and age between updates.
- Popularity is an estimate, so a ranking by popularity is open to debate.
- Career Path shows a career as the player's Wikipedia infobox lists it, and each club with its current crest, even for spells played under an older badge or name.
- Some records are simply wrong. Our editorial policy explains how we correct them, and the contact page is the quickest way to tell us.