Gwangju are the only side in the division who shift their weight backwards when they go behind. It is not that the defenders cannot play forward — there is nobody ahead of them to find. And the players who could stand there are already in this squad.
This document does not propose training methods or ways of running the team. That is the coaching staff's ground. It does one thing: identify a single area for improvement, open up every piece of reasoning behind that judgement, and check whether the players to do it are already in the squad.
The previous analysis asked why Gwangju's channel had closed, and the answer pointed at the defenders. The share of passes carried from the middle third into the final third was the lowest in the league, and the shortfall was concentrated in the defensive line. The observation was correct, but it was a symptom, not an answer.
This time the analysis controls for score state. Gwangju have spent 1,061 minutes behind, far more than anyone else (second: Pohang, 647). What this team does while losing is the season itself. Isolate that phase and the picture inverts.
While behind, the share of middle-third passes that reach the final third is 14.5% for Gwangju and 19.1% for the league — a gap of −4.6pp. That gap splits in two: who receives the ball (distribution) and how much the receiver moves it on (execution).
The distribution effect is effectively zero. Changing who is given the ball will not move this gap. It sits entirely in execution — and broken out by position, it sits in one place.
Gwangju's midfielders progress at 20.1%. The league's do it at 20.3%. Within noise, they are the same.
That matters twice over. It says the midfield rebuilt in the summer window is doing its job — and it says there is almost nothing left to gain by working on the midfield again. The forwards show a wide gap but hold only 6.4% of the ball in this zone, so their effect on the total is small. One place remains.
A gap concentrated in the defensive line does not mean the defenders caused it. There is a clean way to separate the two. When a team goes behind, the opponent sits off to protect the lead — which is why, across the league as a whole, playing forward gets easier the further behind you are. The test is whether that holds at Gwangju.
League defenders, once behind, go into the final third more often (15.4→17.9%) and more successfully (61.5→72.5%). The opponent has dropped off, so the space is there. Gwangju's defenders do the exact opposite — fewer attempts (14.1→11.2%), and the attempts they do make come off less often (63.7→59.7%).
If technique were the problem, completion would move the same way. It moved the opposite way, so the problem is not the pass. It is where the pass has to arrive.
At level score, Gwangju defenders complete 63.7% of their passes into the final third — higher than the league's 61.5%. The same players, with the same technique, invert relative to the league the moment they fall behind. Only one explanation is left.
While behind, Gwangju record 116.1 final-third events per 90. That is last in the league, and the gap to 11th-placed Bucheon (179.5) is larger than the gap from Bucheon all the way up to Seoul in first.
The first thing to suspect is that Gwangju simply have less of the ball. They do not. In the same phase, middle-third activity is 323.0 per 90 against a league average of 348.2 — a difference of 7%. The ball arrives in the middle third at close to league rate, and only beyond it does the number halve.
And decisively, that ratio falls when Gwangju go behind — 0.372 at level to 0.360 trailing. Over the same transition the league lifts it from 0.495 to 0.607.
There is a trap here. Final-third activity is partly the result of the passes we complete. If activity is low because progression is low, the previous chart says the same thing twice.
So the measurement was rebuilt from events that occur independently of our passing: pressing and interceptions in the final third, aerial and ground duels, and fouls won. Every one of them happens while the opponent has the ball or the ball is loose, and none is recorded unless one of our players is standing there. Even with a progression rate of zero, these events occur as long as someone is up the pitch.
The answer is unchanged. Gwangju sit at 20.7 per 90, last of 12, 37% below the league average of 32.8. Because this index is not generated by our own passing, "the final third only looks empty because we never get there" does not explain it.
Across the 12 clubs, the index correlates with progression rate while behind at r = +0.68. The teams with people up there are the teams that play forward.
Take the share of a team's total events that occur in the final third, and measure how that share changes from level to behind. Gwangju's position in this season's K League 1 is unambiguous.
More precisely, Gwangju move backwards. The final-third-to-middle-third ratio above is the other face of the same fact — the league shifts its weight forward while losing; Gwangju hold the ball longer in the middle third. With the centre of gravity unmoved, the only thing that rises is how often the defenders have the ball.
At player level it is sharper still. Take every player in the league who, at level score, does 30% or more of his work in the final third — the players who already stand high. There are 28. 23 of them (82%) go higher when their team falls behind, from an average of 40.4% to 48.6%.
The area to improve is not the defenders' forward pass. It is where the team puts its weight when it is losing.
That is the conclusion. Gwangju's channel is closed not because of the people opening it, but because nobody is standing at the far end of it — and right now this team sends fewer players there when it goes behind, not more.
The target converts through the fit. Lifting the final-third occupancy index from 20.7 to 28.3, the level of 11th-placed Bucheon, predicts a progression rate of 18.1% against today's 14.5%; reaching the league average of 32.8 predicts 19.3%. That is mid-table.
Once the area is set, one question follows: can it be done with the players already at the club?
Start with the end that opens the channel. Group the five Gwangju players with the highest middle-third progression rate while behind and they come to 24.6%. The league average for midfielders in that state is 20.3%. They are better than league average, and most of them are playing now.
| Player | Pos | Level | Behind | Change | Mid-3rd passes | Last 6 rds |
|---|---|---|---|---|---|---|
| Ba Loua (50) | MF | 12.1% | 25.0% | +12.9pp | 80 | 6 apps |
| Ri Yeong-jik (37) | MF | 16.9% | 27.3% | +10.4pp | 144 | 5 apps |
| Mun Min-seo (88) | MF | 14.3% | 23.1% | +8.9pp | 478 | 5 apps |
| Shin Chang-mu (40) | MF | 19.3% | 25.0% | +5.7pp | 193 | 5 apps |
| Ju Se-jong (8) | MF | 25.2% | 25.3% | +0.1pp | 485 | 2 apps |
| Five combined: 24.6% while behind · league midfielders behind: 20.3% · 22.9% of Gwangju's middle-third passes while behind | ||||||
The share is the point. These five account for 22.9% of Gwangju's middle-third passes while behind — effectively identical to their 22.7% at level. The extra volume that arrives when the team goes behind flows to the defenders instead. Gwangju's defenders make +27.0% more middle-third passes per 90 once behind while the midfielders gain only +10.0%; across the league the split is a far more even +23.6% and +17.0%.
João Pedro (17) progresses 20.0% of his middle-third passes at level score. The league average for defenders is 15.4%. His fall to 9.2% while behind cannot be explained by his ability. He has started all six of the last six rounds.
| Player | Pos | Height, level | Height, behind | Change | Last 6 rds |
|---|---|---|---|---|---|
| Kwon Seong-yun (22) | DF | 20.1% | 30.9% | +10.8pp | 5 apps |
| Mun Min-seo (88) | MF | 25.0% | 28.2% | +3.2pp | 5 apps |
| Choi Kyoung-rok (10) | MF | 25.6% | 27.3% | +1.7pp | 4 apps |
| Fridjonsson (11) | FW | 41.7% | 42.9% | +1.2pp | 5 apps |
| Yu Je-ho (14) | MF | 10.2% | 10.0% | −0.2pp | 5 apps |
| João Pedro (17) | DF | 12.9% | 11.7% | −1.2pp | 6 apps |
| Jeong Ji-hun (16) | MF | 35.4% | 29.9% | −5.6pp | 4 apps |
| Shin Chang-mu (40) | MF | 39.5% | 33.8% | −5.7pp | 5 apps |
Height is the share of a player's own events that occur in the final third. Because it is normalised within the player, it is independent of minutes played and simply says where he worked.
Two things show at once. Gwangju do have players who climb when losing — Kwon Seong-yun (22) is a defender and moves up more than anyone at the club (+10.8pp), and Mun Min-seo (88) is the only player who raises both his progression rate and his height. And the two who normally stand highest, Shin Chang-mu (40) and Jeong Ji-hun (16), come down. Shin progresses well from deep once he drops (25.0% while behind) — but every time he drops, one destination disappears.
The strongest evidence that this is achievable is not a league benchmark. It is one of Gwangju's own matches.
Away at Seoul in R14, the goal went in at 48:11 of the second half and the match ran to 96.2 minutes. That leaves 48.2 minutes before and 48.1 after — almost exactly half and half. Same match, same squad, same evening.
Over those 48 minutes Gwangju's final-to-middle ratio was 0.636 — above the league's behind-state average of 0.607. Their progression rate of 28.6% was higher than Gangwon's league-leading 22.3%. The pass-independent occupancy index tripled from 9.3 to 30.0, close to the league average of 32.5.
Honesty requires the other half: the absolute volume fell short. 131.1 final-third events per 90 is 62% of the league's behind-state 211.2. But the ratios and the progression rate cleared league average. And this was the match with the season's highest U23 minutes share (64.5%).
In 42 minutes they produced a league-average centre of gravity. There is no stronger evidence that ability is not the constraint.
The order matters. The goal came first, the weight shift second, the rise in progression last. They did not go forward because the passing worked; the passing worked because they went forward.
Half of deciding what to improve is deciding what not to. The four below look reasonable and are not supported by this data.
| Direction | What the data says |
|---|---|
| Change who distributes | Of the −4.6pp progression gap while behind, the distribution effect is +0.06pp. Changing who receives the ball does not move it. |
| Work on build-up alone | At level score Gwangju's defenders complete 63.7% of passes into the final third against the league's 61.5%. Something that only collapses while behind is not explained by technique. |
| Sign a target man | Since the summer, aerial duels are down 25% and the share of box shots created from crosses has fallen from 35.2% to 19.4%. The team is already not built that way, and there is no evidence for reversing it. |
| Drill final-third patterns | Shots per final-third entry are 0.147, 5th of 12. What happens after arrival is not the problem; arriving is. |
They share a trait. All four try to fix the moment the ball is at someone's feet. What this analysis points at is the other side of it — how many players are standing in the final third while the team is losing.
In one sentence: Gwangju's channel is not blocked — there is simply nobody standing at the end of it. And the players who could stand there are already in this squad. For 48 minutes at Seoul, they did.