Sources
Every rumour is weighted by how often its sources have been right. That is measured from deals that settled, not guessed. A source needs 50 settled calls before it gets a score.
Scored over Summer 2026. The current window is left out until it closes, because the deals that fell through have not settled yet.
Leaderboard
| Source | Tier | Resolved | Hit rate | Brier | Measured trust | vs. tier | |
|---|---|---|---|---|---|---|---|
| 1 | Plettigoal @Plettigoal · Journalist | T2 | 234 | 87% | 0.074 | 85% | ▲ 30 vs tier 2 |
| 2 | FabrizioRomano @FabrizioRomano · Journalist | T1 | 454 | 85% | 0.095 | 81% | ▲ 6 vs tier 1 |
| 3 | David_Ornstein @David_Ornstein · Journalist | T1 | 97 | 81% | 0.101 | 80% | ▲ 5 vs tier 1 |
| 4 | JacobsBen @JacobsBen · Journalist | T2 | 316 | 84% | 0.101 | 80% | ▲ 25 vs tier 2 |
| 5 | MatteMoretto @MatteMoretto · Journalist | T1 | 218 | 74% | 0.138 | 72% | ▼ 3 vs tier 1 |
| 6 | DiMarzio @DiMarzio · Journalist | T2 | 377 | 75% | 0.151 | 70% | ▲ 15 vs tier 2 |
| 7 | TEAMtalk @TEAMtalk · Aggregator | T3 | 353 | 52% | 0.244 | 51% | ▲ 11 vs tier 3 |
| 8 | MirrorFootball @MirrorFootball · Aggregator | T4 | 78 | 62% | 0.294 | 41% | ▲ 21 vs tier 4 |
| — | FabriceHawkins @FabriceHawkins · Journalist | T2 | 42of 50 | — | — | not yet measured | — |
| — | Arsenal (official) @arsenalofficial · ClubMedia | T2 | 0of 50 | — | — | not yet measured | — |
| — | Deadline Desk @deadlinedeskhq · Journalist | T2 | 0of 50 | — | — | not yet measured | — |
| — | Touchline Wire @touchlinewire · Journalist | T2 | 0of 50 | — | — | not yet measured | — |
Hit rate counts advancing reports on deals that happened, plus denials on deals that did not. Calling a deal off counts the same as calling it on. Brier also rewards being confident when right, and it is what the odds use.
Calibration
Does “Probable” actually mean probable?
read 7 days before each outcome · updated 22h agoMeasured at the probability 7 days before each rumour settled, not the final reading. A signing sits at 95% the day before it is announced, so scoring against that would prove nothing. If a band stops meaning what it says, it shows here.
Scores come from public posts and public outcomes. They rate a set of reports, not the person who wrote them.
One known problem: a loan can show as no transfer when squad data lists the parent club. That gives the reporter a miss they did not earn, and it hits whoever covers loans most. We check and correct these after each window closes, so read a figure as a floor, not a verdict.