Strongest at Home,
Unable to Win There

Gimpo FC began 2026 without a home. The pitch at the Solter Football Stadium was being completely re-turfed, so from R2 to R14 they played 13 consecutive away matches. The home opener came on 7 June, R15 against Jeonnam.

Three months on the road went well. Across those 13 matches: 6 wins, 5 draws, 2 defeats, 23 points — 1.77 per match. They came home in fifth, in the play-off places.

What came after is the problem. From R15, 12 matches: 2 wins, 6 draws, 4 defeats, 12 points — 1.00 per match. In the September table Gimpo are 7th.

The easy explanation would be “coming home broke their rhythm”. The data says something else. Gimpo play the best home football in the league. They just do not win.

Re-read the matches by shot quality and Gimpo’s 10 home games are the best home games in K League 2. Non-penalty expected goal difference per match (chances created minus chances conceded, by xG) is +0.61 — 1st of 17.

At home they create 1.53 xG a match and concede 0.93. Replaying the same shots at their xG gives 1.78 expected points per home match, also 1st.

Away they are ordinary: +0.05 per match over 15 away games, mid-table. Gimpo’s good football is concentrated at home.

Yet the 10 home matches have brought 11 points, 1.10 per match, 11th in the league — 6.8 fewer than expected. Replay those 10 matches 20,000 times with the same shots and 11 points or fewer comes up 3.4% of the time.

Away it is the other way round: +1.7 against expectation. Over the season Gimpo are 5.1 points short of expected (−5.1, the 6th-largest shortfall in the league) — and all of it comes from home matches.

The two right-hand columns are the key (penalties and own goals excluded). Away, Gimpo scored as expected and conceded as expected. At home they scored 5.3 fewer than expected and conceded 4.7 more. Both ends went wrong at once.

Between 18 July and 16 August Gimpo played five home matches. Four of them finished 1-1. The other (R21, Chungbuk Cheongju) was a 3-0 win.

Gimpo controlled all four. Shots 79 : 34, non-penalty xG 8.18 : 3.44. The four matches were worth 8.8 expected points and brought 4. Replay them with the same shots and 4 points or fewer comes up 4.1% of the time.

Look at the opponents’ goals. Yongin took 9 shots (non-penalty xG 0.47 in total) and scored with one worth 0.05. Gyeongnam took 9 (0.61) and scored with one worth 0.07; Cheonan took 6 (0.60) and scored with one worth 0.17. Shots that go in perhaps once in ten attempts went in, in matches where almost nothing was conceded. Gimpo, meanwhile, shot 8.2 xG across the four matches and scored 4.

Four in a row is a lot to call luck alone. There is no evidence for a single cause either. This is as far as the data goes — Gimpo’s home draws did not come from being outplayed.

Gimpo’s attacks start the moment they win the ball. Attacks that begin with a defensive regain come 54.7 times per 90, 1st in the league.

And they are quick. xG created within 15 seconds of winning the ball is 0.40 per 90, 1st; within 10 seconds it is 0.27, 2nd. 62.8% of Gimpo’s non-penalty xG comes from attacks that start with a regain — also 1st. In a test that shuffles 10-second transition xG across the league’s team-matches, a figure as high as Gimpo’s came up 2.6% of the time.

Where do they win it? Split the pitch into nine zones and Gimpo win the ball back in the middle third 23.3 times per 90 — the most in the league. They are above average in their own third (24.3, 4th) and in the final third (4.9, 4th) too. The attacks that follow are big ones — 3rd for xG from own-third regains, 3rd for xG from final-third regains.

The last row is Gimpo’s blank. Attacks from set pieces rank 13th for xG and have produced 1 goal all season. Set-piece shots: 39, worth 3.5 xG — one goal or fewer from that comes up 11% of the time.

⭐ At home, though, this weapon carries less of the load. The share of non-penalty xG from regain-started attacks is 79.0% away and 65.0% at home — a gap of 13.9 points, the largest of the 17 clubs. At home, chances built by moving the ball against a settled defence are worth 0.48 xG a match, more than double the away figure (0.19). Fast attacks have not disappeared (15-second transition xG per match: home 0.53 · away 0.40). A team that wins it and finishes it away also has to carry the slow attack at home.

Gimpo’s shots gather in the middle. Of their non-penalty shots, 42.3% come from the central box (inside the width of the goal area, outside the 6-yard box) — 1st in the league (average 33.0%). But one step further in — the 6-yard box, right in front of goal — they take only 3.4% of their shots, the fewest in the league (average 8.2%).

The central box is where they shoot most and score least. Goals there are 3.7 below xG, 15th; conversion is 11.3%, 13th.

Split home and away and the story sharpens. They took 62 central-box shots in each. Away they scored as expected (+0.9); at home, xG 9.6 produced 5 goals. Of the home scoring shortfall (5.3 non-penalty goals), 4.6 comes from this one zone. Home 6-yard-box shots in 10 matches: 2 (away: 8).

Who took them? The two players who carried the central box away from home stalled at home.

On target, too. 31.1% of Gimpo’s shots hit the target, 14th in the league. Take only the shots on target and recompute expected goals from which cell of the goal they went to (top/bottom × left/centre/right) — away, 49 on target produced 16 goals (placement-expected 13.8); at home, 43 produced 10 (placement-expected 12.6). At home, opposing goalkeepers stopped 2.6 more than their placement would suggest.

I split the point where each attack first enters the final third into five lanes (wide · half-space · centre). Like everyone else, Gimpo enter most often down the flanks. The difference is in what happens next.

Entries through the half-spaces end in a shot 31.9% of the time (5th). Entries down the wings: 17.3% (13th). Gimpo’s good chances come not from the wing but from the channel between the wing and the centre.

From wide, they cross a lot: 18.2 per 90, 3rd, with 21.2% completed (13th). But goals from cross-assisted shots are 2.1 below xG — 16th. Shots from set pieces are 2.5 below, 16th. Attacks that end in the air have not paid off (2 headed goals, both away).

“Conceding more” does not mean a weak defence. Gimpo are one of the stingiest sides in K League 2.

Shots conceded per match 9.7, xG conceded 1.06 — both the 2nd-fewest in the league. Opponents move from midfield into the final third 42.2% of the time, the 2nd-lowest, and xG conceded inside the box is 0.79 a match, the 2nd-fewest.

The method is clear. Split the point where opponents enter Gimpo’s final third by lane: entries through the centre are 11.5%, the lowest in the league (average 14.5%), and entries through the half-spaces are 28.4%, the highest. Close the middle, push them outside. The few central entries that get through are small too: 0.024 xG per entry, the 3rd-smallest.

They concede little and still let in more. Open-play goals conceded exceed xG by +0.22 per 90 — the 2nd-largest excess in the league. Non-penalty goals conceded: 29, from opponent shots worth 24.2 xG.

The breach is right in front of goal. 10.0% of shots conceded come from the 6-yard box, 4th — few shots overall, but the closest ones get through. Goals conceded from the central box are 2.8 above xG, 3rd. Goals from cross-assisted shots are 3.4 above xG (3rd). Of the 29 non-penalty goals conceded, 8 came from the 6-yard box and 7 from shots worth less than 0.1 xG.

The paradox is sharpest when the score is level. Level, Gimpo hold opponents’ progression (midfield → final third 38.8%), shots after entry (21.6%) and 10-second counters after a turnover (xG 0.08 per 90) all to the lowest in the league. In the same state, open-play goals conceded exceed xG by +0.40 per 90 — the largest excess in the league. Goals conceded while level: 16, from shots worth 11.5 xG.

What happened when opponents’ shots were on target? The portal records which cell of the goal each on-target shot went to — two rows (top, bottom) × three columns (left, centre, right).

First I took, across all 212 league matches, the share of on-target shots in each cell that became goals. Top corners 43–47%, bottom corners 29–37%, centre 19% (top) and 13% (bottom). That gives a “placement expectation” — how many goals shots into those cells produce on average — to compare with the goals actually conceded.

Gimpo faced 74 non-penalty shots on target, the 3rd-fewest in the league. They conceded 29; the placement expectation is 21.9. The gap of +7.1 is the 2nd-largest in the league. Opponents did not pick the corners unusually well — the placement-expected save rate is 70.4%, about league average — but the actual save rate was 60.8%.

⭐ This is what separates home and away. At home: 29 shots on target, 14 goals, placement expectation 8.9 — a gap of +5.1. Away: 45 on target, 15 goals, expected 13.0 (+2.0). The home excess in goals conceded seen earlier (+4.7, penalties and own goals excluded) comes almost entirely from this one step.

Two goalkeepers shared the season.

Bosang Yoon’s spell overlaps with the summer after the return home. Of the 23 on-target shots he faced at home, 12 went in, against a placement expectation of 7.0. Jeonghyeon Son faced 29 away, conceding 11 (expected 8.4).

For a side 5th on expected goal difference, Gimpo spend more time behind (25%) than in front (21%). They have drawn 11 times — 4.5 more than the 6.5 their shot quality suggests, the 3rd-largest excess in the league.

• Leading — they shoot, but it does not go in. Entry → shot 32.4%, 3rd; 10- and 15-second transition xG both 2nd. Yet open-play goals are 0.46 below xG per 90 (15th). While leading: 4 goals from 6.3 xG. Points dropped from winning positions: 12.
• Level — the tightest defence, and the most conceded. The paradox from the previous section lives here. In attack, regain chains are 58.2 per 90, 1st.
• Trailing — progression works, dead balls do not. Build-ups reach midfield 61.3% of the time, 17th (lowest), but midfield → final third is 60.8%, 3rd. Set-piece xG is 0.06 per 90, 17th (lowest). Points per match after conceding first: 0.45, 14th.

Gimpo’s shots come from their foreign forwards. 56.7% of the team’s shots and 61.5% of non-penalty goals belong to foreign players — both 1st in the league. Their share of minutes (28.0%) is 3rd, but their share of output is the biggest.

Luis Mina was the centre of it. 81 shots (27% of the team’s), 10 goals (3 penalties), xG 11.9. With him on the pitch the team created 1.16 xG per 90; without him, 0.72. His last appearance was R22 (16 August), after which he moved to Suwon Samsung. There have been only 4 matches without him, so this piece does not judge the effect of the transfer.

The squad is the oldest in the league: minutes-weighted average age 30.5, 1st. Substitutes have scored 3 goals, 10% of the team’s total, 16th. Red cards: 5, most in the league (average 1.3).

Beyond home and away, I cut the matches several other ways to see what else splits Gimpo’s results.

① Opponent strength. Split opponents into the September top six and the rest, and the home shortfall has an address. In 7 home matches against the other ten clubs, Gimpo took 7 points — expected: 14.4. They dominated those matches, 1.57 : 0.70 in non-penalty xG. Away against the same clubs, they took +5.2 more than expected.

② Time. Cut matches into 15-minute blocks and the last block at home stands out: 6 goals conceded at home after the 75th minute, from 2.9 xG. In the same block away, Gimpo scored 6 and conceded 2. Gimpo have the oldest squad in the league (weighted average 30.5) and rank 16th for goals by substitutes — but nothing in the data links these to the late home goals.

③ Who plays together. In the 1160 minutes with four foreign players on the pitch at once, Gimpo’s non-penalty xG difference was +0.66 per 90; with three (934 minutes), −0.03. Gimpo’s attack is complete only when all four are there.

④ Signals inside matches. Across the 25 matches I checked rank correlations between match stats and “points − expected points”. The more they crossed (ρ −0.61) and the more possession they had (ρ −0.55), the less they took relative to expectation. ⚠ Teams that are behind keep the ball and cross more — score effects are mixed into these correlations, so they are not read as causes. 25 matches is a small sample.

⑤ Discipline. 5 red cards, most in the league; 11.3 fouls per 90, 3rd. In R16 against Chungnam Asan, a first-half red card was followed by two goals conceded and a defeat.

The whole breakdown on one page: two strengths, and five risks that swing results.

In the end Gimpo’s story is this. Away, they are a team that wins it and finishes it; at home, they become a team that has to push the ball into the central box against a settled defence. And at home the last cell in front of both goals went wrong at once — Gimpo’s shots did not go in from the central box, and opponents’ shots on target went in more often than their placement suggests.

1–3 are the results, 4–5 are where those results were decided, and 6–8 are how Gimpo control matches.

Representative matches are chosen by the series’ four rules — ① the match closest to the season average ② the most Gimpo-like match (shots per entry) ③ the match where the result most outran the process ④ the match where the process most outran the result. ① and ④ picked the same match (R15, Jeonnam), so ① moves to the next in line.

③ was away, ④ was the homecoming. The rules picked Gimpo’s two faces on their own.

Gimpo play the best home football in K League 2. And at home they have taken 6.8 points fewer than expected. They were not outplayed; the last cell in front of both goals went wrong at the same time.

Of the remaining seven matches (R28–R34), 6 are at the Solter Football Stadium. This season will be decided at home. Three things to watch:

• The central box at home — xG 9.6, 5 goals there so far. Do they start taking the extra step into the 6-yard box?
• The last cell on target — home goals conceded from shots on target are +5.1 against placement expectation. Do balls into the same cells start being stopped at the league-average rate?
• An answer for the settled attack — at home, where weaker sides sit deep, what does a team built to win it and finish it do with the ball? Matches without Luis Mina so far: 4.