What two Premier League midfielders reveal about performance data, tactical responsibilities and recruitment decisions.
By Tobias Schweinsteiger
(Head of Football at GAMECODE.Ai | Former professional footballer and coach at FC Bayern München II, Hamburger SV, 1. FC Nürnberg and VfL Osnabrück | Bundesliga and Champions League analyst for DAZN)
Last season, Declan Rice ranked in the 98th percentile for defensive performance among central and defensive midfielders across Europe’s five major leagues. James Garner ranked in the 37th percentile, placing 370th out of 580 central midfielders (see the percentile explanation below). Eight matches into the new season, the two are much closer: Rice ranks 150th and Garner 180th out of 463 central midfielders.
Having coached, I know that players do not suddenly become completely different footballers over the summer. Their form, responsibilities and the team around them can change considerably. If I were recruiting either midfielder, I would want to understand why the numbers now look so different.
Declan Rice of England looks on during a FIFA World Cup 26 match (Photo by Chris Arjoon/Icon Sportswire)
Across the full 2025/26 Premier League season, GAMECODE.Ai ranked Rice in the 98th percentile overall (7th out of 580 central and defensive midfielders across Europe’s top five leagues). Garner ranked in the 83rd percentile overall (100th). Offensively, the gap was narrower: Rice ranked in the 96th percentile (23rd), compared with Garner in the 90th (60th). Defensively, Rice led by a clear margin, ranking in the 98th percentile (7th), compared with Garner in the 37th (370th). Rice was stronger across every defensive category, from central areas and the penalty box to wide spaces and transitions. Over a full season, he was clearly the more complete midfielder.
Before assessing Garner’s defensive ability, however, I would consider the context. Everton conceded 532 opposition shots with Garner on the pitch; Arsenal conceded 273 with Rice playing. These are team figures, not individual mistakes or achievements. They show how different the defensive environments were, but do not explain the entire performance gap.
What a coach sees before the ball is lost
Imagine your team attacking with both full-backs high. Your number ten receives the ball between the opposition lines, while the holding midfielder supports possession from behind. Where should he position himself? That depends on the centre-backs, the position of the far-side full-back and whether teammates can press immediately if possession is lost. Move too far forward and an important passing lane is exposed. Stay too deep and he may become disconnected from the attack. The decisive defensive decision can come several seconds before the opponent wins the ball.
When comparing Rice and Garner, I would examine similar situations. Where were they positioned before possession changed? Did they recognise the most dangerous passing option? Did they step towards the ball or protect the centre? Could they delay the counterattack long enough for teammates to recover?
Rice’s full-season defensive profile makes him the stronger candidate for a role requiring extensive defensive coverage. With Garner, I would investigate whether difficult situations resulted from his individual decisions or Everton’s collective structure. A midfielder who makes more tackles is not necessarily the better defender. Fewer interventions can reflect excellent positioning or simply a more organised team. Match footage helps distinguish between those possibilities.
This becomes particularly relevant when examining 2026/27. After eight Premier League appearances, Rice’s defensive percentile has fallen from 98 to 66, while Garner’s has risen from 37 to 61. The second radar chart now suggests that they are almost equal defensively. I would not draw that conclusion from 712 minutes for Rice and 662 for Garner. The reference group has also changed, from 580 midfielders to 463. These percentiles describe relative performance within different populations; they do not directly measure individual improvement or decline. I would examine whether their responsibilities have changed, how their teams defend transitions and whether particular opponents influenced the early-season results. The full season provides stronger evidence. The opening eight matches raise questions worth investigating.
Different qualities, different requirements
In possession, Rice’s advantage is particularly interesting. Across the opening eight matches of 2026/27, he recorded 134 actions that enabled a teammate’s subsequent progressive action, compared with 83 for Garner. That equates to 16.9 versus 11.3 per 90 minutes. Rice also led in our midfield-specific expected threat assist metric, recording 3.39 compared with Garner’s 2.53 per 90 minutes. These figures interest me because good midfield play is not only about spectacular passes. Receiving under pressure, drawing an opponent towards you and creating space for a teammate can be just as valuable.
Rice contributes to those situations more frequently. His four assists in the early-season sample, compared with none for Garner, reinforce that impression, although assists alone are not sufficient evidence. Garner brings other qualities. In 2025/26, he delivered 50 wide crosses, compared with Rice’s 33. Both recorded almost identical numbers of attacking-half set pieces: 191 for Garner and 195 for Rice. Garner also registered 15 shots following free kicks, with seven on target, compared with seven attempts and three on target for Rice.
That does not automatically make Garner the more accurate crosser or the more effective free-kick taker. It does, however, identify strengths worth examining.
Imagine two recruitment briefs. The first team wants to dominate possession, commit its full-backs forward and protect large spaces when opponents break through the press. Rice’s defensive versatility and contribution to progression make him the stronger candidate.
The second team already has reliable defensive cover and central progression but needs greater quality in wide delivery and set-piece involvement. Garner’s profile could be relevant, particularly if the transfer budget is more limited. That is why I believe the footballing requirements must come before the statistics.
Edin Terzić and his coaching staff developed 144 individual metrics across more than 20 categories, based on their own philosophy. At GAMECODE.Ai, we helped translate those requirements into a measurable framework. Different coaches expect different decisions from their midfielders. A coach who prioritises aggressive counter-pressing may value different qualities from one focused on protecting central spaces after possession is lost. Data helps us examine those differences, provided we understand what we are measuring.
Rice’s published market valuation is €120 million, compared with €35 million for Garner. Market valuations reflect more than performance, but Rice’s full-season record provides compelling evidence of his superior all-round quality. Which statistical picture would I trust? I would start with the complete season. Rice’s performances provide substantially stronger evidence of his overall contribution. Garner’s improved early-season defensive ranking is interesting, but eight appearances are not enough to overturn that assessment.
I would then consider what my team actually needs. Before recommending either player, I would want to watch the moments that define central midfield: when the full-backs advance, possession changes and the midfielder must recognise danger before it becomes obvious.
That is where I would make my judgement. Not from one radar chart, and certainly not from eight matches alone.
Data note: Analysis by the GAMECODE.Ai Bot of Premier League performances during 2025/26 and the opening eight appearances of 2026/27. Percentiles compare central and defensive midfielders across Europe’s top five leagues (580 players in 2025/26; 463 in 2026/27). Opposition figures represent team outcomes while the player was on the pitch. Market valuations are dated Transfermarkt estimates.
Percentile calculation
A percentile (also known as a percentile rank) divides a distribution into 100 equal parts. It indicates the proportion of observations at or below a given value. For example, at the 95th percentile, 95% of observations are less than or equal to the value at that percentile.
Leave a Reply