Once Caught, Never Ahead Again

Bucheon took the lead in 14 of their 29 matches. In half of them — seven — that lead was wiped out, and not once did they go back in front afterwards: 0 wins, 5 draws, 2 defeats. The seven leads that survived are the season’s seven wins.

Six of the seven wins came the same way: score first, never concede an equaliser. The one exception is the opening day (R1, 3–2 away at Jeonbuk). Bucheon were behind twice, at 0–1 and 1–2, and won it with a penalty seven minutes into stoppage time. It is their only comeback win of the season, and the only time they were ahead in that match was right at the end.

So the whole season hangs on the first goal.

Before answering, the baseline needs setting. 32 points and tenth place are not a straight reading of this team’s level.

Across 29 matches, expected goals for are 27.5 and expected goals against 45.8. Expected goal difference −18.3; actual −7. That is 11.3 goals gained on expectation, and 7.8 of them at the defensive end.

A defensive overperformance of 7.8 goals (11.5 on the 28 observed matches) is large. If it is skill, Bucheon are doing something xG cannot capture. If it is variance, it will be gone next season. The two readings point in opposite directions, so I tested it three ways.

One — it is not the goalkeeper

Two goalkeepers shared the season. If one man were performing above himself, the two should look different.

Their save rates are 2.1 points apart. The excess did not split by save rate but in proportion to how much xGA each faced (31.63 → +8.63; 12.88 → +2.88). Against the opposing goalkeepers it is 70.8% vs 68.8% — a 2.0-point edge is nowhere near big enough to produce 11.5 goals.

Two — it is not the defensive structure either

The control group is FC Anyang, parsed through the same pipeline: same league, same season, 29 matches.

Anyang block 2.0 points more, and the on-target gap is only 4.2 points. Yet the defensive overperformance is +11.5 for Bucheon and −3.3 for Anyang. If almost the same suppression produces opposite results, suppression cannot be the cause.

Three — a variance test, and the missing match

I treated each of the 408 opposition shots as a coin flip weighted by its xG and simulated the season 200,000 times.

On 28 matches it is narrowly significant, p = 0.026. But those 28 matches are the sample with the 0–5 defeat taken out. Put R27 back under any reasonable assumption and p rises to 0.07–0.11. The defence looks good partly because its worst match is missing, and this test was exactly the one most exposed to that bias.

In short, the repeatable part is at most around four goals; the other seven or so cannot be separated from variance with this data. The safer reading is that Bucheon are weaker than 32 points. Which leaves one question — why is the attack only this good?

Once in the final third, Bucheon shoot readily — 13.21 shots per 100 final-third passes, the highest of the four groups compared (opponents 11.12, Anyang 11.27). The problem is how often they arrive: 22.2 vs 33.2 successful entries a match, 67% of the opposition’s figure.

A 51.1% success rate is lower not only than Bucheon’s opponents (60.8%) but also than Anyang (56.6%) and Anyang’s opponents (62.9%). Split the 9.8-point gap by method and more than half of it comes from one cell.

Going long is not the weakness. Long-pass success is level with the opposition, 40.5% vs 41.7%. What Bucheon do not use is the middle distance — and long passes (43.5%) and free kicks (12.6%, 2.1 times the opposition’s share) fill the space it leaves.

How rarely they play the mid-range pass

By lane, the centre stands out. Bucheon complete 32.8% of their central entries; their opponents 48.0%, Anyang 46.0%, Anyang’s opponents 44.7%. Bucheon alone are in the thirties; the other three sit between 44 and 48%.

It looks blocked. Open up how the ball goes in, and a different picture appears.

That single line is the whole section. When Bucheon play a short pass into the centre, it gets through as often as anyone else’s. The centre is not closed. They simply use that route for only 17.2% of central entries (opponents 22.2%). The rest is long passes (42.0%) and free kicks (18.9%), which come off one time in five.

Mid-range passes into the centre also split, 36.5% vs 61.6% — a 25.1-point gap. The cell from the previous section turns up again here.

Bucheon’s entries lean left: 44.8% (opponents 38.7%). So do 47.3% of their final-third passes and 52.1% of their crosses. It is tempting to read the bias itself as the problem. The output says otherwise.

Bucheon turn wide entries into shots more often than their opponents do — 16.7% vs 12.3% on the left, 16.0% vs 13.8% on the right. Going left is not the wrong choice.

The loss is in the centre. Successful central entries number 78 against 160 — half — and the xG they produce is 1.31 vs 3.46. A 2.15 difference from a single lane.

Look at who plays the mid-range forward pass, and two things appear at once.

And the players whose passes arrive are the ones who try least. The share of these passes that actually reach the final third: Ahn Tae-hyun 41.4%, Kim Jong-woo 41.2%, Basani 38.7% — from only 29, 34 and 31 attempts. Baek Dong-gyu, by contrast, tries 102 and reaches the final third 3 times (2.9%).

One player makes the left

Count touches on the left of the final third and Gallego has 1.9 times the next player. He spends a third of his touches in that zone.

If the attack cannot be built through entries, the other route is winning the ball back. Bucheon do win it more often — 72.8 recoveries a match against 59.0. But they win it in a different place.

The share of recoveries made in the final third is 5.7% vs 9.2%. They win plenty, mostly in their own half, so less than half as many recoveries turn into attacks.

And the few that do end from further out

Split shots by what happened just before them, and one cell stands out.

The recovery spot is almost the same — the median X of the defensive action before those shots is 75.2 vs 76.9. They win the ball in the same place, but only Bucheon shoot from further out (median distance 20.1 m vs 18.7 m; 61.0% vs 53.3% from beyond 18 m).

The same finger points at the same place here too — within three seconds of winning the ball, Bucheon choose a mid-range forward pass 2.4% vs 3.9% of the time.

Everything above comes together in xG by game state.

As ratios: leading 0.54 · level 0.53 · trailing 0.78. They drop off when ahead or level, and only come out when behind. And the default state of a match is level — which makes 0.53 this team’s baseline.

Here is the story laid over real matches rather than averages. They were chosen by the numbers, not by impression — ① the win closest to the season average on a composite measure, ② the widest entry gap, ③ the widest xG gap, ④ the most shots per final-third pass.

Five hypotheses raised during this analysis did not survive the data. All five were mine. They stay in — a rejected hypothesis marks the edges of the conclusions that did survive.

The diagnosis has narrowed to “they do not use the route into the centre”, so that is where intervention starts. The order below is lowest cost first — not signings or a new formation, but changing the choices of players already in the squad.