xgEdge

Hidden gems

A player who won his place three weeks ago is still being read at the rate of a man who was not playing, and that is the gap this page is for. Every row is cheap, owned by few, and has a rising minutes trend off rounds that have actually been played. The argument for him is printed beside him rather than left implied.

The board's best qualifier

Lewis Dunk

DEF · BHA · 4.5m · owned by 1.5%

12.4

Next 5

His last 4 rounds average 89.8 minutes against 52.5 before them, a gain of 37.3.

GW11 · Next 5 gameweeks · Data updated 19 Nov 16:00

Lewis Dunk is the best-projected gem on the board: 4.5m, owned by 1.5%, and 12.4 projected points over the next 5 rounds.

His last 4 rounds average 89.8 minutes against 52.5 before them, a gain of 37.3. He has started his last 4 rounds in a row. He is owned by 1.5% of the game. He costs 4.5m, under the 5.5m ceiling this list is set to. The model rates his next 5 fixtures 2.2 out of five on average.

Price

4.5m

What he costs today, against the 5.5m ceiling every row here had to clear.

Next 5

12.4

His projected points over the next 5 rounds, added up. The qualifiers are ranked on their ranking score, which counts nearer rounds more.

Own

1.5%

The share of managers holding him, against the 10% gate this board opens at.

What the gates turned away

17 other cheap, low-owned players are left off: over the last 8 stored rounds their minutes are not rising, or their expected involvements are falling.

Position
More filters (1 set)

1 filter active Clear filters

Showing 2 of 2 players · 2 of 502 match your filters 2 of 2 players 2 of 502 match your filters

1 Lewis Dunk DEF BHA HUL (A) Compare 12.4 Next 5 4.5m 1.5% owned Why he's here
  • +37.3 mins
  • xGI/90 level
  • 4 starts in a row
  • 1.5% owned
  • 4.5m
  • Run 2.2/5
More
Pts
2.9
xGI
0.02
Mins
749
xG/90
0.00
xA/90
0.02
Clean sheet
38%
Exp. minutes
69
Over replacement
−4.2
Ranking score
8.54
Next fixtures
  1. GW 11 HUL (A) 2.9
  2. GW 12 NEW (H) 2.8
  3. GW 13 BOU (A) 2.0
  4. GW 14 NFO (A) 2.1
  5. GW 15 EVE (H) 2.6
2 Costinha DEF BHA HUL (A) Compare 6.6 Next 5 4.5m 0.0% owned Why he's here
  • +52.5 mins
  • xGI/90 unreadable
  • 2 starts in a row
  • 0.0% owned
  • 4.5m
  • Run 2.2/5
More
Pts
1.5
xGI
0.25
Mins
354
xG/90
0.08
xA/90
0.17
Clean sheet
38%
Exp. minutes
34
Over replacement
−10.0
Ranking score
4.53
Next fixtures
  1. GW 11 HUL (A) 1.5
  2. GW 12 NEW (H) 1.5
  3. GW 13 BOU (A) 1.0
  4. GW 14 NFO (A) 1.1
  5. GW 15 EVE (H) 1.4
each row's projected points over the rounds ahead, drawn to scale every bar on one axis of 0 to 20

What qualifies a gem, and what kind of figure each of these is

The trends are arithmetic over matches that have already been played: the last 8 stored rounds, split into the most recent 4 and whatever came before them. Nothing about them is projected, which makes them the most checkable figures on this page and the least predictive. A rising trend is evidence that something has changed, not a forecast that it continues.

“Cheap” is the ceiling you set — 5.5m as this board stands — and “few” is under 10% of the game. Both are controls rather than definitions, because the slot a reader is filling is not something this page knows.

Every row says why he is here in the same six lines: the minutes trend, the expected goal involvements per 90 trend, the starts in a row, the ownership, the price and the fixture rating. A line is highlighted where it is an argument for him: starts in a row from 2, and a run rated 2.5 or easier out of five. Every row has an improvement signal, because rising minutes is the gate; a cheap, low-owned player whose minutes are flat is not on the list.

The minutes gate is a gain of 22.5 minutes a round on the earlier half, or an average of 60 minutes for a player with nothing behind his recent rounds at all. Below that a player is being rotated rather than promoted.

The expected involvements are Fantasy's own per-round figures, read per 90 over each half and only where each half holds 90 minutes. A falling one leaves a player off the list; one that cannot be read is said so on the row rather than counted either way.

A player the game has published no ownership figure for is left off rather than counted as owned by nobody. A player this app holds no stored rounds for is left off rather than credited with a trend. Missing is not nought.

Every ranking here is on projected points for the round ahead or on the ranking score for the rounds ahead, which are the units this model is measured in. The value column is those figures restated over a price — it is a way of scanning the board, and nothing on this site backtests it.

Bonus points are projected nowhere on this site, so they are in none of these figures.

The board is every player the game would let you field, the flagged included. Whether a doubtful player is worth picking is a judgement the projection makes with his published chance of playing, not one made by hiding him.

What we've tested

  • Backtested

    Projected points

    xG Edge error
    1.617
    Simple baseline
    1.842

    Average error in points per player, across 456 players. The baseline predicts each player's average so far; the test replayed gameweek 2 after it was played, recorded Sunday, 6 September 2026.