Publishing a ranked board the day before a round is a testable claim. Gameweek 1 is scored, so here is the test, run against the snapshot committed on the morning of 21 August, before a ball was kicked.
229 rated players started or came on. Sorted into six groups by where they sat on the board that morning, this is what they returned.
| Board position | Players | Mean GW1 points | Mean price |
|---|---|---|---|
| 1st to 47th | 39 | 4.08 | £6.7m |
| 52nd to 112th | 39 | 3.92 | £5.6m |
| 113th to 180th | 39 | 3.44 | £5.7m |
| 181st to 244th | 39 | 2.69 | £5.2m |
| 245th to 304th | 39 | 2.72 | £5.3m |
| 305th to 369th | 34 | 2.06 | £5.3m |
The gradient runs the right way from top to bottom, with one inversion in the middle. The top group returned almost exactly double the bottom group.
That is the flattering version. The rank-to-points correlation across all 229 is -0.198, which is a real relationship and a weak one. One gameweek of football is a tiny sample and the honest summary is a nudge in the right direction rather than a crystal ball.
The obvious objection
Look at the price column. The top group averages £6.7m and the bottom £5.3m, and expensive players score more. A board that ranks Haaland above a £4.0m Hull defender has not demonstrated anything a price list could not.
So the real test is whether the board separates players who cost the same. Splitting each price bracket into its better-rated half and its worse-rated half:
| Price bracket | Players | Better-rated half | Worse-rated half | Gap |
|---|---|---|---|---|
| £4.0m to £4.5m | 43 | 3.52 | 2.57 | +0.95 |
| £5.0m to £5.5m | 111 | 2.95 | 2.69 | +0.25 |
| £6.0m to £6.5m | 53 | 4.65 | 2.31 | +2.35 |
| £7.0m and above | 22 | 3.64 | 5.64 | -2.00 |
Three brackets out of four went the right way. The £6.0m to £6.5m bracket is the standout, where the better-rated half more than doubled the worse-rated half over fifty-three players.
The £7.0m bracket went backwards.
What happened at the top
Twenty-two players priced at £7.0m or above took the pitch, and the ones our board rated lower outscored the ones it rated higher by two points a head.
The reason is not mysterious. Haaland sits 2nd on the board and Bruno Fernandes 3rd, the two highest-rated outfield players in the game, and they returned 2 points each. Cole Palmer returned 13 and João Pedro 11, and neither is anywhere near the top of the premium rankings.
Twenty-two players is not a sample anyone should draw a conclusion from, and one blank from Haaland is not evidence about Haaland. But it is worth saying out loud rather than burying under the three brackets that worked.
The pattern by position
| Position | Top 50 rated | Everyone else |
|---|---|---|
| Goalkeeper | 4.20 | 3.09 |
| Defender | 3.75 | 3.00 |
| Midfielder | 4.57 | 3.12 |
| Forward | 3.50 | 2.36 |
All four positions went the right way, with midfield the widest gap at nearly a point and a half.
What to do with this
The useful conclusion is about where to spend attention rather than whether the board is good.
At the premium end, everyone already knows. Haaland is 68.8% owned and Bruno is 50.1%, the whole game has read the same information, and a rating adds almost nothing to a decision that thousands of people have already priced. Gameweek 1 made that point rather bluntly.
The £6.0m to £6.5m bracket is where it earned its keep, and that is exactly where a manager has the most decisions and the least information. That bracket is four or five slots in most squads, it changes most often, and it is where nobody can tell you with confidence who is worth owning.
So the transfer worth thinking hardest about this week is not the captain. It is whichever £6.0m midfielder or defender you picked on a hunch, because that is the part of the squad where a rank actually separated players over the opening weekend.
The caveats, stated plainly
This is one gameweek and 229 players, at a correlation of -0.198. Four of the ten fixtures involved a promoted club whose squad the board could not rate at all, and every one of those players is absent from the tables above, because there was no pre-deadline rank to sort them by.
This will be worth rerunning at ten gameweeks and worth trusting at twenty. It is published now because a board that only gets marked when it looks good is not a board worth reading.
Every figure comes from the committed snapshot of 21 August for the ranks, and today’s snapshot for the returns.