Elite 10k Analysis

Effective ownership (EO) counts a player once for every manager who starts him, twice for every manager who captains him and three times for a triple captain (TC), divided by the number of managers in the tier - so a heavily captained player can go past 100%. Own is the share with him anywhere in their 15, XI the share starting him, and the captain and TC shares are the armband picks as the managers made them. FPL only shows picks after the deadline, and the multipliers it gives back are the ones that counted once auto-subs had been applied. Each gameweek uses its latest capture.

Managers are sampled in rank bands - 1 to 1,000, 1,001 to 10,000 and 10,001 to 100,000 - and each sampled manager is weighted by how many managers his band holds, so every share stands for the whole tier rather than for the sample. The bands, the numbers sampled and the weights are printed below. The top 10k is ranks 1 to 10,000 at the capture; the top 100k adds the band above it. A tier is shown only when every band inside it was sampled - sampling of the 10,001 to 100,000 band began on 13 September 2026 - and is otherwise marked unavailable rather than estimated from the bands that were.

The same-manager change is measured only over managers who appear in both this gameweek's capture and the previous one, weighted the same way: their level now minus their level a week ago. It is free of any churn in who was sampled, but the sample is redrawn each week, so the overlap is usually around fifty managers and the figure carries roughly plus or minus 14 points. It is shown as a dash when there is no previous capture, and for a tier whose overlap misses one of its rank bands - normally the case for the top 100k.

The model column is Sideline Quant's own projection for the gameweek, taken from the last run before the deadline and summed across a double gameweek, so it is what the model said in advance and never a hindsight number. Our rank is the player's place among every player in that run, and the chance of 10+ is that run's probability of a ten-point haul, which is absent for a double gameweek and for the pre-season bundle that did not record it. Likely starters are players the run gave more than 60 expected minutes. Gameweek 1 predates the retention of runs and uses the committed pre-season bundle instead.

The headline change compares this gameweek's level with last gameweek's: two separate samples of about a thousand managers, one taken each week, so it carries sampling noise of roughly plus or minus 4 points at 95% on a share near 30%, and a change smaller than that is noise rather than a move. It shows the template moving rather than the same managers trading. The same-manager change in the full table is that narrower reading. A player last week's sample held and this week's does not sits at the foot of the full table at 0.0; where there is no previous capture, both readings are shown as a dash.

Over-owned: of the twenty highest-EO players, the eight the model ranks lowest by expected points. Under-owned: of the model's twenty highest expected-points likely starters (expected minutes over 60), the eight with the lowest effective ownership in this tier.

Once a gameweek settles, every row carries the player's realised FPL points, summed across his fixtures that gameweek and before any armband, and each list carries the mean of the rows that scored. The lists are fixed by the ownership and the projection at the deadline; the points are filled in afterwards and nothing is re-picked. Until the gameweek settles, points and means are shown as a dash and the previous gameweek's verdict is shown instead, if that gameweek has settled.