Why "Role Watch" Is the Real Edge in FPL

A role is a cause, not a sample

That shrinkage is correct for noise. It's badly wrong for a genuine role change, because a role isn't a statistical sample that needs to converge. It's a cause. If you can see that a player now starts attacks deeper, finishes them higher and has a new teammate covering the space behind him, you don't need eight matches of xG to confirm the mechanism. You need the eye test to confirm it once, and the numbers to confirm it isn't a mirage.

Our own projection engine has now moved Groß to roughly 0.54 xGI per 90 across the next six gameweeks, which is most of the way to his current output. It got there, but look at what the market charged anyone who waited for it:

The elite were early: 44.6% of the top 1,000 already owned him by the GW2 deadline, while he was still £5.5m. But even they hedged. Only 26.1% started him that week, which means roughly two in five elite owners watched his 13 points from the bench. By GW4, 73.5% owned him and he'd started to rise. He's now had three price rises in ten days and more than 670,000 net transfers in this gameweek alone. That last wave isn't Role Watch. It's the algorithm arriving.

I. Initial phase vs terminal phase

Track where a player starts an attack and where he finishes it. In Phase 1 (first-line build-up), does the midfielder drop between the centre-backs to collect from the goalkeeper? In Phase 3 (box entry), where is he when the cut-back or the cross arrives?

A player who drops deep in Phase 1 is easily dismissed as "a defensive midfielder". But if the same player sprints 40 yards to become the third man crashing the penalty spot in Phase 3, his ceiling is far higher than his heat map suggests. Understat gives you the cleanest single number for this split:

That's the whole Role Watch thesis in six rows. Groß's involvement used to end in build-up: 83% of the xG he touched came from possessions where someone else took the shot or played the final pass. This season it's 41%, and his threat per 90 has more than doubled. At the other end of the table, Rice's build-up share has risen to 96% and his threat has more than halved. Same player, same shirt, a very different job.

II. Positional freedom and half-space hunting

Is the player pinned to a structured lane, or does he have licence to roam? Wingers told to hold extreme width to stretch a low block pile up touches, but mostly low-value touches, and crosses into a packed box rarely turn into big xA. Players given freedom to drift into the gap between the opposition full-back and centre-back operate in the most dangerous zone on the pitch. That's where the 0.35+ xG chances and the cut-backs come from.

You can see the difference in the quality of the chances a player creates, not just the quantity:

Fernandes has created nearly as many chances as Groß and fewer than half the expected assists. That isn't a criticism of the most reliable points-scorer in the game; it is simply a reflection of where his chances come from. For Role Watch, the question is always which kind of creator a role change is turning someone into.

III. The anchor effect: teammate dependency

A player's fantasy ceiling is rarely set in isolation. It is usually unlocked (or capped) by the player standing fifteen yards behind him.

When a team runs a lightweight, dual-roaming double pivot, both midfielders have to temper their forward runs to protect the rest-defence: someone has to stay home when the ball is lost. The moment a dedicated, physical holding midfielder enters the XI, the other one is released. His job description changes from "one of two who might go" to "the one who goes".

That makes the team sheet a leading indicator in its own right. When you're watching an attacking midfielder, you're also watching the player behind him. Our Predicted Lineups page is the quickest way to see whether the anchor is expected to start before the deadline.

• →Anchor comes in → the #8 is freed: expect more box arrivals, higher shot volume, a falling build-up share.
• →Anchor goes out (injury, rotation, suspension) → the #8 is pulled back: expect the reverse, often within a single match.
• →Anchor is an FPL asset too: since 2025/26, midfielders earn 2 points for 12 defensive actions in a match. Bentancur is averaging 14.59 per 90 with four defensive-contribution points already; Xhaka is at 10.8 with four.

IV. Counter-press proximity

When the ball is turned over, where is your asset standing? If he's the first line of the counter-press in the final third, he benefits from transition returns: winning the ball back 25 yards from goal against a defence that's still facing the wrong way is one of the highest-converting situations in modern football.

The tell in the data is an attacking player with an unusually high defensive workload and a low build-up share. Tavernier fits the profile: 8.92 defensive actions per 90 (more than many holding midfielders), goals in three of his five matches, and the lowest build-up share of anyone on our shortlist below. He isn't building attacks; he's finishing them, often after winning the ball back himself.

V. The hierarchy behind the ball

The fifth pillar isn't about movement. It's about pecking order: who takes the penalties, the corners and the direct free-kicks, and what happens to that order when a teammate comes off, gets injured or is sold. Set-piece hierarchies change far more often than the templates notice, and they change silently: nobody publishes a press release when the left-footed corner taker is benched.

It also works as a check on everything above. Groß is listed as Brighton's primary penalty taker, yet his 0.98 xG this season is identical to his non-penalty xG: every bit of it has come from open play. The penalty is upside his underlying numbers haven't started counting. For a primary taker like Saka, always check how much of the xG is non-penalty before calling a spike a role change.

Fluid 6–8–10 mobility

Groß isn't playing one position. In the 6 phase he drops deep to start possession from the centre-backs, using his passing range to bypass the first pressing wave. In the 8 phase, as the ball moves forward, he slides diagonally into the right half-space and finds the pockets between the opposition's midfield and defensive lines. In the 10 phase, when the ball reaches the final third, he arrives in the box as a primary receiver, occupying positions you'd normally associate with a second striker.

 HOW GROSS MOVES THROUGH A BRIGHTON ATTACK
 │
 ├── PHASE 1 · BUILD-UP (the 6)
 │     Drops beside the centre-backs to collect and beat the first press
 │
 ├── PHASE 2 · PROGRESSION (the 8)
 │     Slides diagonally into the right half-space, between the lines
 │
 ├── PHASE 3 · TERMINAL (the 10)
 │     Arrives at the penalty spot or the cut-back zone as a receiver
 │
 └── WHAT MAKES IT SAFE
       A fixed #6 who does NOT follow the ball forward
       Andrés: 10.73 defensive actions per 90 so far

The data footprint

The numbers back up every stage of that:

Twice as many shots, far more of the dangerous passes, and less than half of his involvement now ending in build-up. That's the signature of a player whose job has moved up the pitch. Through five gameweeks, his 2.33 xA is the most of any regular midfielder or defender in the league (360+ minutes), his rise in xGI per 90 is the largest of any of them with a meaningful sample last season, and his 47 points put him top of all 199 midfielders. None of it has come from penalties.

The anchor: an honest correction

The neat version of this story goes: Brighton found a proper holding midfielder, the holding midfielder freed Groß, and Groß exploded. The minutes don't quite allow it:

Groß's surge started before Chema Andrés arrived. His 13-pointer at Chelsea came alongside Malick Yalcouyé, and his 17 at Coventry came with Yasin Ayari and Yalcouyé starting and Andrés on for the last 25 minutes (he assisted). Brighton have rotated the players around Groß almost every week, with Diego Gómez the only other constant. Whatever released Groß was in place from the opening weeks. That's exactly why the managers who read the role, rather than the teamsheet, were in at £5.5m.

So where does Andrés fit? He's the evidence that the role is sustainable. On his first start against Arsenal in GW5 he played 84 minutes, scored, and crucially for Groß, looked like the fixed #6 the structure needs. His 10.73 defensive actions per 90 are the profile of a genuine anchor: disciplined, positional, the man who cleans up second balls and stays home when the ball is lost. Groß scored 14 points in that match.

Here's the Role Watch edge that's still live. Our model projects Andrés for just 3.52 points over the next six gameweeks, meaning it doesn't yet expect him to be a regular starter. If he does keep the shirt, the structure protecting Groß's forward role is more secure than the projections assume. If he drops out, watch Groß's box arrivals closely over the next two matches before you panic, because the surge began without him.

Where the numbers are now

Our engine now projects Groß for 28.16 points over GW6–11. That's fifth among all midfielders, behind Fernandes (£11.9m), Saka (£9.5m), Semenyo (£8.4m) and Mbeumo (£7.9m). At £5.8m, he's the cheapest of the five by more than £2m. The algorithm has caught up. The window for buying him as a differential has closed; he's now template, which makes him a shield rather than an edge (see the effective ownership section of The Elite Edge).

Brighton had a second role story, and the elite read it even faster. Maxim De Cuyper is a left-back whose numbers look like a winger's: 1.48 xG and 1.16 xA from defence, alongside just 3.4 defensive actions per 90 because he spends very little of his time defending. Top-1,000 ownership jumped from 34.4% at the GW2 deadline to 67.4% at GW3, and he's already risen four times, to £4.9m.

The decoy winger vs the terminal winger

Watch the two flanks against each other. Very often one winger is an isolator, hugging the touchline to draw a double team, while the winger on the opposite side is told to come inside and attack the back post. The isolator racks up crosses that get headed clear; the back-post winger racks up tap-ins. On the FPL site their basic numbers can look almost identical until you look at where their shots come from. The tell: the terminal winger has a low build-up share, high shot volume and high xG per shot. The decoy has the reverse.

The inverted full-back push

When a full-back inverts into central midfield in possession, look at what happens to the interior midfielder on his side. He's usually pushed higher, into the forward line, as an extra attacker. Tracking a full-back's inversion is often the quickest way to predict an attacking midfielder's goal surge well before the midfielder's own numbers move.

Gvardiol is the model inverted defender in the data: 88% of the xG he touches comes from build-up, the profile of a player who spends his possessions in midfield rather than on the overlap. Brighton's Ferdi Kadıoğlu is the one to watch now: his involvement so far is modest (0.25 xGChain per 90), and a jump in his build-up volume would be the early warning that Brighton are using him to push a midfielder higher.

The auxiliary winger at full-back

De Cuyper's archetype: a full-back or wing-back whose side attacks down his flank and trusts the rest of the back line to cover. The signature is a defender with meaningful xG and a low defensive-action count. He's the most efficient asset type in FPL when it works, because he collects attacking returns at defender prices and still has a clean-sheet floor. The risk is the mirror image: when the manager decides he's defending too little, the role can be switched off in a single team talk.

The anchor as an asset

Since defensive contribution points arrived, the holding midfielder isn't just the man who frees your attacker; he can score on his own terms. Bentancur and Xhaka both have four defensive-contribution points already. Xhaka adds a creative layer on top (1.40 xA), which is why a £5.5m holding midfielder has outscored several £7m attackers. Role Watch cuts both ways here: when a team drops a deep playmaker for a pure destroyer, the destroyer's defensive-contribution numbers often move before anyone's projections do.

The pulled-back playmaker

The most expensive archetype to miss: an attacking midfielder whose side has lost its anchor, or changed shape, so he now drops to collect the ball. His touches go up, his heat map looks busy and his name still carries its old price. His build-up share climbs and his threat falls. That's the next section.

📌 Understat, 2025/26 vs 2026/27 through GW5. Threat per 90 = non-penalty xG + xA per 90.

📌 xGI/90 last = Understat non-penalty xG + xA per 90 in 2025/26; 'now' = FPL xG + xA per 90 through GW5. Build-up = this season's build-up share. Top-1k = share of the top 1,000 overall owning him at the GW4 deadline ('-' = unowned or below capture threshold). xP 6GW = FPL King projection, GW6–11. Davis and Mitchell are defenders.

The seven checks

• →Has the personnel around him changed? Check whether an anchor has come in or gone out, and whether he's expected to start.
• →Where does he register touches in Phase 3? Build-up involvement is only a problem if it stops him arriving in the box. Build-up share tells you which.
• →Are the returns open play or dead ball? Compare npxG with xG, and check xA per key pass.
• →Is the physical demand sustainable? Look at the substitution pattern around 60–70 minutes, and at the model's expected minutes.
• →Has the market already paid for it? Check Price Change Predictions and the elite ownership trends: under ~20% top-1,000 ownership, you're early; over ~60%, it's the template.
• →Does the fixture suit the role? A terminal #8 against a low block is a different asset from the same player against a high line: check the attacking side of Fixture Difficulty.
• →Write down what would prove you wrong. If the anchor is dropped or the box arrivals disappear for two matches, the thesis is dead: sell on the role, not on last week's points.

FPL King

Prediction Engine Team

Written by the team behind FPL King's prediction engine, turning bookmaker odds, xG modelling and squad optimisation into plain-English advice.