How Often Do Transfer Rumours Come True?
Every Premier League transfer rumour we record is checked against what actually happened. This page is the result, recalculated every night. It is the one question about transfer news that nobody publishes an answer to - so we did.
The finding
About 76% of Premier League transfer rumours never happen.
Of 232 rumours linking a player to a Premier League club, 55 ended in a confirmed transfer to that club - 24%. Give or take about 7 percentage points on this sample size.
A second independent slice agrees. Narrow it to the 132 rumours the BBC's gossip column attributes to a named outlet, a different sample, gathered a different way - and 24% came true. Two separate cuts of the data landing in the same place is the best evidence we have that the figure is real rather than an artefact of how we asked the question.
The detail
Turned around: when a Premier League transfer was confirmed, we already had a rumour on record linking that player to that club 33% of the time - a median of 4.6 days before the club announced it. That counts only rumours we recorded before the announcement (5 arrived after the deal was already done, and are excluded), so nothing here is credit for a rumour written up after the fact.
Of the confirmed moves we recorded a type for, 61% were permanent transfers and 39% were loans.
How this is calculated
- A rumour is a link, not an article. One player linked to one club counts once, however many outlets report it. We hold 2,002 press mentions, 1,173 of which name a destination club, producing 426 distinct rumours.
- "Came true" means confirmed afterwards. The transfer must be officially confirmed to the club the rumour named, and confirmed after we first recorded the rumour. A deal announced before we saw any rumour is not something we predicted, so it does not count.
- Confirmations come from the Premier League, not from us. We read each club's official Transfer Centre feed. That is also this study's main limitation: a move to a club outside the Premier League is much harder for us to verify, so the headline figure counts only rumours with a Premier League destination (232 of 426). Across all destinations the rate is 21%, but that blends a population we can check with one we largely cannot, so we do not lead with it - and we draw no conclusion about whether rumours abroad are more or less reliable, because our own coverage gap would explain any difference.
- Recent rumours are included. A rumour reported last week has had little time to happen. Excluding everything under 30 days old moves the rate by about a point, so we leave them in rather than pick a window that flatters us.
- We use the median, not the average, for timing. A few long-running sagas pull the mean well above the typical case.
- What we deliberately do not publish. We do not rank outlets by accuracy: we tested it, and with the sample available the differences between them are indistinguishable from chance, so any table would be inventing a league from noise. We do not publish transfer fees, which are recorded for almost none of these deals. And we do not report how "hot" a rumour was before it completed, because we changed how that score is calculated on 29 July 2026 and the series either side is not comparable.
- Free transfers and releases. The Premier League feed does not distinguish a free transfer from a fee-paying one, so "permanent" includes frees. Released players are not counted as transfers at all, which means a free agent who later signs elsewhere is missing from the confirmed set.
- Period covered. Rumours recorded from 6 Jun 2026 onward, across 351 confirmed transfers.
Updated 24 Oct 2026. Recalculated nightly.
Citing this
Free to quote with a link back. Suggested wording:
Only 24% of Premier League transfer rumours end in a confirmed transfer to the club named, according to Transfer Scout, which tracked 232 rumours against official club announcements.
Happy to share the underlying data or answer questions on the method. The figures update every night, so this page is always the current version.