Chungbuk Cheongju FC have the second-highest share of the ball in the league and allow their opponents few chances. Yet they have conceded 11 more goals than their opponents' expected goals — by far the biggest gap in the league. I put all 212 league matches on one scale and opened the attacking, defensive, transition and set-piece phases one by one to find where those eleven goals came from.
On 18 April, in R8, Chungbuk Cheongju pinned Gimhae back at home. They entered the final third 30 times and piled up shots; non-penalty xG read 2.67 to 0.55. They went ahead on 79 minutes. In the sixth minute of stoppage time one Gimhae shot went in. Its chance of scoring was 27%. 1-1.
Up to that match Chungbuk Cheongju had not won in eight. Their first win only came on 7 June, in R15 away at Seoul E-Land. First 13 matches: 0 wins, 10 draws, 3 defeats. Re-take every shot 20,000 times and those 13 matches are worth 18.9 expected points; they took 10. In the September table they are 5W 14D 7L, 11th on 29 points. Their 14 draws are the most in the league.
One question, then: why does a team that keeps the ball and gives little away concede so much? I open the four phases in turn — attack, defensive block, transition, set pieces — and then look at the last yard in front of goal. The answer sits not in the phases but in that last yard.
I unrolled all 26 matches minute by minute — when Chungbuk Cheongju led, when it was level, when they trailed, and the moments the goals went in.
Chungbuk Cheongju spent 62% of their minutes level, 17% ahead and 20% behind. They went ahead in 14 matches and conceded while ahead in 11 of them (2nd in the league). Of the 41 goals conceded, 10 came from the 85th minute on and 8 from the 90th. R8 Gimhae (90+6), R11 Gimpo (90+2), R13 Hwaseong (90+11), R20 Suwon Samsung (90+6) — the goals that changed results are bunched in stoppage time.
Add up the expected goals of every shot opponents took against Chungbuk Cheongju and you get 30.0. The goals that actually went in: 41. The difference is +11.0.
That is the largest gap among the 17 clubs. Next are Yongin (+3.7); the league average is −0.7. For everyone else expected and actual goals against agree to within a few goals. And the expected goals Chungbuk Cheongju allow are not high — the 6th-lowest in the league. They have not conceded a single penalty (the only club in the league). All eleven goals came from open play or set pieces.
A gap like this could be chance. But it held for all 26 matches — 20 conceded in the first 13 (14.6 expected), 21 in the 13 after (15.4 expected). So which phase is leaking? I start with the attack and work through.
Chungbuk Cheongju keep the ball. Their pass share of 54.5% is 2nd in the league and their pass completion of 82.0% is the highest. Their 62.7 balls lost per match are the fewest in the league.
But the ball does not travel forward easily. Settled attacks reach the final third 9.3 times per match — the fewest in the league. Once there, they do get the shot away: 31.3% of settled entries end in a shot, 2nd in the league. From winning the ball to the first shot takes 44 seconds on average, the slowest in the league. This is a team that resets and builds again rather than breaks.
The result is open-play xG of 0.81 per 90 — 11th, mid-table. Turning those chances into goals fell slightly short (goals minus xG per 90 −0.10, 14th). Much of the scoring rests on one man: García scored 13 of the team's 31 goals.
The defensive block is solid. Opponents reach Chungbuk Cheongju's final third 27.8 times per 90, the 2nd-fewest in the league. They get into the box 11.5 times per match, the 3rd-fewest, and their xG from settled attacks is the 4th-lowest.
The gap is at the last step. Opponents who reach the box go on to shoot 53.0% of the time, the 4th-highest in the league. And Chungbuk Cheongju concede 1.23 open-play goals per 90 — the 2nd-most in the league. The chances they allow (xG) are mid-table; the goals they concede are near the top. The block is not the front end of the problem.
Their reaction when the ball is lost is good too. They win it back within five seconds 26.7% of the time, 2nd in the league. So the xG they allow within ten seconds of losing the ball is 0.28 per match, the 5th-lowest in the league.
The other direction is weaker. They win the ball by a defensive action 53.2 times per match (16th), and the xG they create within 15 seconds of a regain is 12th. That is natural for a side that keeps the ball — the opponents have it less, so there is less to win back. But the goals they concede within 15 seconds of losing the ball number 11, 6th in the league. Little xG allowed, above-average goals conceded — the same pattern again.
Set-piece xG created is 0.15 per 90 (7th) and allowed 0.15 (8th lowest) — mid-table at both ends. Set-piece goals scored fell short of expectation (0.10 per 90, 14th) and set-piece goals conceded were close to it. They have won 6 penalties, 2nd in the league, and conceded none. The eleven goals did not come from set pieces.
None of the four phases explains the eleven goals. One place is left: after the shot has headed for goal.
Of the 87 shots on target against Chungbuk Cheongju, 41 went in. A save rate of 52.9% is the lowest in the league, against a league average of 69.6%. Put the other way, 41.4% of shots on target became goals — far clear of second-placed Seoul E-Land (32.5%). Using the six goal cells the portal records, I calculated how many goals shots on target into those cells produce on average (placement expectation): 28.6. 41 went in — +12.4, again the worst in the league.
Where did they leak? In the six-yard box they actually conceded fewer than expected (9 from xG 10.7). The leak is beyond it. Central box: 85 shots faced, 19 conceded (xG 12.3, +6.7 — the largest excess in the league). Outside the box: 80 shots, 7 conceded (xG 3.2, +3.8 — again the largest excess). From shots with less than a 10% chance of scoring alone they conceded 18, the most in the league (average 8.5).
Four goalkeepers have played this season: Donggeon Noh (R1–R3), Sunghoon Cho (R4–R17), Seunghwan Lee (R18–R24) and Sungbin Cho (R25–R27). All four conceded more than the placement expectation.
Why this cannot simply be pinned on the goalkeepers · All four pointing the same way can also mean it is not one man's problem. Chungbuk Cheongju's defenders block 16.1% of shots, the 3rd-lowest in the league — fewer blocks in front means cleaner shots for the keeper. The placement expectation knows nothing of shot speed, screening just before the shot, or deflections. What this data says is "on-target shots into the same cells went in far more often than the league average" — no more.
When did the extra goals go in? Split by match status, the answer is sharp.
At level Chungbuk Cheongju conceded 16 from 16.3 expected — exactly what was expected. Ahead is different. In the 455 minutes they led, opponents created only 5.9 xG, and Chungbuk Cheongju conceded 15. Their goals-minus-expected while leading (+9.1) is the worst in the league. They failed to hold the lead in 11 of the 14 matches they led (2nd in the league), dropping 20 points (2nd). Their 7 draws after scoring first are the most in the league.
Seongnam, in episode 22, also fell apart when ahead. But the two are different. Seongnam dropped deep the moment they led, allowed a lot and conceded what they allowed (13 from 12.6 expected while ahead). Chungbuk Cheongju do not allow much even when ahead: 1.18 expected goals against per 90 while leading, about 61% of Seongnam's (1.94) and the 6th-lowest in the league. Yet those chances go in. On 9 October, in R28, the two meet in Cheongju.
The mark the leaked goals left on the table is the draw. Chungbuk Cheongju drew 14 times. Replay the same shots and the expected number of draws is 7.0 — an excess of 7.0, the largest in the league.
Of the 14, 7 came after they scored first, 4 after they fell behind and 3 were goalless. More of them were wins turned into draws than defeats rescued.
The weight fell early in the season. The first 13 matches went 0W 10D 3L: 18.9 expected points, 10 taken. Replay those 13 matches with the same shots and the chance of taking 10 points or fewer is only 2%. The 13 matches since the first win in R15: 18.5 expected, 19 taken — almost identical. Yet by goals the two halves are nearly the same — 15 scored and 20 conceded in the first 13, 16 and 21 in the 13 after — and both conceded more than expected (14.6 · 15.4 expected). What differed was the order and timing of the goals: early on they turned wins into draws (10 draws); later the draws fell to 4.
Home and away split too. At home: 20.2 expected points, 12 taken. Away: 17.3 expected, 17 taken. Almost all the dropped points were dropped in Cheongju.
All 26 matches side by side in non-penalty xG, split into Edge, Even and Outplayed at an xG difference of ±0.3.
In the 8 matches where they had the edge, Chungbuk Cheongju went only 3W 5D 0L. Even: 1W 6D 2L from 9. Outplayed: 1W 3D 5L from 9. The 5 matches they had the edge in but did not win (R4 Ansan 1-1 · R7 Cheonan 2-2 · R8 Gimhae 1-1 · R10 Paju 0-0 · R23 Gyeongnam 1-1) are this season's loss.
1–5 are the last yard in front of goal, 6–8 its consequences, 9–10 the shape of the team.
Representative matches are chosen by this series' four rules — ① the match closest to the season average ② the most characteristic (shots per final-third entry) ③ result furthest ahead of process ④ process furthest ahead of result.
A 0-0 away. Keep the ball, allow little, score little. One of the fourteen draws.
Scored on 28 and 52 minutes, conceded on 76 and in the sixth minute of stoppage time. The two goals conceded came from shots worth 0.44 xG in total.
The first win, thirteen matches in. This time Chungbuk Cheongju were the ones scoring on 85 and 90+1.
xG 2.67 : 0.55, ahead on 79 minutes. In stoppage time a shot with a 27% chance went in.
② and ④ are this piece in two matches. In both, the result was changed by one low-chance shot in stoppage time while they were ahead.
Opponent final-third entries are the 2nd-fewest in the league and expected goals against the 6th-lowest. They did not allow a lot; a lot of what they allowed went in.
All four goalkeepers conceded more than the placement expectation, and the share of shots blocked by defenders is the 3rd-lowest in the league. The last yard in front of goal is made by the keeper and the defence in front of him together.
Set-piece xG and goals conceded are mid-table, and they have not conceded a single penalty.
They did concede a lot from the 90th minute on (8). But opponents did not create much in that time — expected goals against after the 75th minute rank 9th. They did not collapse late; the few late chances went in.
Partly. But it showed up for all 26 matches, with four different goalkeepers, in the first 13 matches and in the 13 after. Luck that keeps falling the same way makes you suspect a structure.
Chungbuk Cheongju keep the ball, win it back quickly when they lose it, and give opponents little. No phase has collapsed badly enough to explain eleven goals. Yet in the last yard in front of goal they conceded 11 more than expected, many of them in stoppage time while ahead. And so wins became draws — fourteen times.
The remaining fixtures: R28 Seongnam (H, 9 Oct) · R29 Daegu (A, 17 Oct) · R30 Yongin (H, 25 Oct) · R31 Paju (H, 1 Nov) · R33 Busan (A, 21 Nov) · R34 Suwon FC (A, 29 Nov). First up are Seongnam from episode 22. Both struggle to hold a lead — but Seongnam concede what they allow, and Chungbuk Cheongju concede more than they allow. Three things to watch.
Limits · The portal's xG model is not published. Expected points treat shots as independent. Placement expectation uses only the six goal cells and knows nothing of shot speed, screening or deflections. Goalkeeper figures are approximations by starting keeper. Conceding more than expected mixes goalkeeping, defending just before the shot, and luck; this data cannot separate the three. The figures were found while scanning many measures (multiple comparisons).