WHAT THE DATA SAYS · 01

K League 2 2026 · Hwaseong FC · 23 matches · 33,522 events

Hwaseong are not a side that holds the ball. They are a side that spends it.

They attempt the second-fewest passes in the league and get more box shots out of each one than anybody. Their long-pass share is third-highest while their duel count is the lowest and their offside count the highest — the ball goes behind the line, not into a body. And the football only completes itself after the break: 86 shots in first halves, 154 in second, with possession unchanged.

Han HyukjinSeptember 2026Data: K League Data PortalWhat the data says — part 1

The conclusions first

They do not hold the ball. 401.7 passes per match, 16th of 17; 45.7% possession, under 50% in 18 of 23 games. They are 5th in the table.
Each pass is worth more. 2.08 box shots per 100 passes, among the league's best; 70.8% of their shots come from inside the area, the highest in the league.
Long, but into space rather than into a body. 3rd for long-pass share, last for duel volume, first for offsides.
They become a different side after the break. 86 first-half shots become 154; possession goes 45.3% to 46.0%. They shoot more in second halves even when leading at the interval.
Ten scored and one conceded between the 46th and 60th; eight conceded after the 76th. The window they surge in is the window that opens.
Possession does not explain them. Correlation with points −0.05, and the sign flips using first-half possession. They can win without the ball; they do not need to be without it.

1The 5th-placed side that passes least

How much ball you have and how much you make of it are different questions

Hwaseong attempt 401.7 passes per match across 23 games — 16th of 17. They sit 5th in the table, and their completion rate of 83.9% ranks 8th. This is not a side that cannot pass; it is a side that chooses not to.

So divide the passes by what they produce: box shots per 100 passes. Hwaseong return 2.08, the highest in the league — though Daegu at 2.07 and Paju at 2.06 make that effectively a three-way tie. The point is not the rank but the position: near the bottom for volume, at the top for yield.

Figure 1 · Volume against yield. Passes attempted per match on the x-axis, box shots per 100 passes on the y. Dashed lines are league means. Hwaseong sit top-left — they circulate little and produce a lot. Bottom-right (Seoul E, Ansan) is the opposite. Paju share the quadrant but complete only 73.1% of their passes, the league's worst, so the same position means something different. Hover a point for the club's numbers.

2Long, but into space rather than into a body

The usual side-effect of playing direct is missing here

Long passes make up 13.4% of Hwaseong's distribution, third-highest in the league. Sides like this normally rack up aerial duels — you hit a target and fight for the second ball. Paju, who top the long-pass share, contest 60.6 duels per match, the most in the league.

Hwaseong contest 40.5, the fewest of all 17. What spikes instead is elsewhere: 2.43 offsides per match, the most in the league, against 1.0 for their opponents in the same fixtures. The average position of those offsides is X 75.2 in attacking direction — right on the opponent's last line.

Long passes many, physical contests fewest in the league, offsides most in the league. Three numbers telling one story — the ball is not played into a body, it is played behind.

Figure 2 · Sides that fight, sides that run. Duels per match on the x-axis, offsides per match on the y. Hwaseong sit alone in the top-left. Paju are bottom-right — also a long-ball side, but one that resolves things through contact. This is an association, not a cause: a high offside count signals frequent runs in behind, not successful ones.

3No cheap shots

Ordinary shot volume, extraordinary shot location

Their 11.8 shots per match rank 9th — dead average. But 70.8% of them come from inside the box, the highest in the league and clear of Yongin (69.6%) and Suwon FC (64.8%). Inverted: only 3.3 shots per match from outside the area, 16th of 17. There is no "just hit it" in this side's vocabulary.

Figure 3 · Share of shots taken inside the box. Expected goals per shot are 0.124 for Hwaseong and 0.120 for their opponents — barely different. The difference is in the distribution of where they shoot from. On the portal's official model Hwaseong post 30.22 xG, 6th in the league, with goals/xG at 1.03: they have scored what they should have, not run ahead of it.

4This team becomes a different side after the break

Not more of the ball — different things done with the same amount

Across 23 matches Hwaseong take 86 shots in first halves and 154 in second halves — a 79% increase. The first thing to check is possession, and it barely moves: 45.3% to 46.0%. They are not getting more of the ball.

Figure 4 · First half to second. Totals across 23 matches, indexed to the first half = 100. Passes fall 7% and completion drops 4.9 points, while passes into the final third rise 39%, crosses 46% and shots 79%. Accuracy sold, penetration bought — that is the second half in one line. Their opponents, over the same split, go from 119 shots to 128, a rise of 8%.

It is not game state

The obvious objection is that they chase games. Split by the score at half-time:

Figure 5 · Second-half shots by half-time state. 6.3 when leading, 6.7 when level, 7.1 when behind. All three are more than 1.7× the first-half average of 3.7 (red dashed line). They shoot more after the break even when ahead. Game state amplifies the effect; it does not create it. What does create it cannot be settled from this data — their substitutions average the 70th minute, later than their opponents' 67th.

5The clock of this team

The hour after the break is theirs; the last quarter is the price

Figure 6 · xG and goals across six bands. Hwaseong left, opponents right; the two figures on the far right are actual goals for and against. 46-60 min: ten scored, one conceded — 28% of the season's goals inside fifteen minutes. 0-15 min: 40 opponent shots against 26 of their own. They failed to score in the first half of 12 of 23 matches. 76-90+: eight conceded, a third of the season's total. The window in which they surge (9.11 xG) is the window that opens for the opposition too (7.87 xG).

Those eight goals after the 76th minute may be a structural cost of this football, or a conditioning and rotation problem. This data cannot separate the two. Doing so would need physical data — distance covered, sprints — which the portal does not carry.

6In from the flanks, finished through the middle

The route in and the place it ends are not the same place

Split the 1,018 passes into the final third by channel: 41% left, 17% centre, 42% right. Their opponents put 23% through the middle in the same phase. Hwaseong do not enter centrally. 83% of entries go down a channel. The goals, though, cluster in the opposite place — eight from the central area in front of goal, four from the penalty-spot area, three from inside the six-yard box.

Figure 7 · Entry channels and finishing zones. The figures on the left are the channel split of passes into the final third (n=1,018); the bubbles on the right are the shot zones of the 31 open-play goals, sized by count. Zone names follow the official match record (GA = goal area, PA = penalty area, AK = arc). Read alongside the fact that 53% of their shots arrive within 15 seconds of winning the ball, this wide route looks less like slow circulation and more like fast delivery.

7Where the 36 goals came from

Not a set-piece side, and heavily reliant on one player

Figure 8 · Goal types and involvements. Twenty-three of the goals (64%) came from an open-play assist. Hwaseong win 5.3 corners a match, 3rd in the league, and have scored three from them, with none direct from a free kick — the cheapest available improvement sits here. Below, goal involvements: Plana alone was involved in 18 of the 36 (50%). Petrov has seven from 1,074 minutes across 18 games — about an hour a match.

8Possession does not explain this team

Strip out the reverse causation and the conclusion gets careful

In the 35-41% possession band Hwaseong average 2.00 points per match; above 48%, 1.67. Read plainly that says "the less ball, the better." It should not be read that way. Matches like R20 against Daegu — 35% possession, won 5-1 — are cases where the lead produced the low possession, not the other way round.

Take first-half possession, which carries less game state, and the sign flips: 2.00 points per match at 45% or above, 1.36 below it. Across all 23 matches the correlation between possession and points is −0.05, effectively zero. The only quantity that tracks points is xG (+0.54).

Figure 9 · Possession against points. Full-match possession on the left, first-half possession on the right; the horizontal bars are mean points per band. Same side, same 23 matches, opposite conclusions depending on which possession figure you use. The accurate sentence is this: Hwaseong can win without the ball; they do not need to be without it to win. With n=23, each band holds six to twelve matches. This figure points a hypothesis; it does not test one.

9The season has been a convergence on itself

Span	Pts/match	Possession	Passes/match	Direct%	GF	GA
R1-R12	1.58	47.5%	456	47%	1.33	1.08
R13-R24	1.82	43.6%	388	42%	1.82	1.00

As the season has gone on they have taken less of the ball, passed less, and scored more — and picked up more points. With eleven or twelve matches on each side of the split this is too early to call a trend. But it points one way.

10Four matches that show this football best

Chosen by summing eight identity axes as z-scores

The 23 matches were standardised and summed across eight axes: ① low possession ② few passes ③ directness (share of shot-producing possessions taking three passes or fewer) ④ share of entries down the channels ⑤ share of shots in the second half ⑥ offsides ⑦ share of shots inside the box ⑧ mean pass length. The index is my own operational definition, not an official metric.

R20

The model at full power

5-1H · Daegu

Five goals off 268 passes, the fewest of the season. Average pass 20.8m, 22% of passes over 30m, completion 75% — the trade of accuracy for penetration at its most extreme. 70% of the possessions that ended in a shot took three passes or fewer, and the second goal went from their own half straight to the striker's feet. Plana scored four. The opponent was Daegu, 4th in the table and the league's highest-volume shooting side at 14.2 per match; they were held to nine.

Watch for — How five goals come out of 35% of the ball. Count the steps from recovery to finish.

35%Possession

268Passes

70%≤3-pass shot phases

1.99xG

R24

The purest Hwaseong win

2-1H · CB Cheongju

Directness 83% — the season's highest. Most shot-producing possessions ended within three passes and 85% of the shots came from inside the box. Three shots in the first half, nine in the second. Both goals came after the break, the winner at 90+3 from a single pass out of midfield. Four offsides means the space behind was attacked again and again.

Watch for — The most recent match and the finished article. The patient first half and the second-half switch in one game.

35%Possession

83%Directness (season high)

85%Shots inside box

3 → 9Shots 1H → 2H

R12

The flank machine and what it costs

3-2H · Busan

The season's highest xG at 2.63. Fourteen crosses, and 87% of the entries into the final third came down the channels. The second goal repeated the R20 shape — one pass out of their own half, finished at the edge of the box. The same match also exposes the cost: two goals conceded in six minutes either side of the half-hour, and 16 shots worth 2.07 xG allowed.

Watch for — The repeated supply route from flank to box — and the counter-attacking exposure that shape creates.

2.63xG (season high)

87%Wide entries

14Crosses

2.07xG against

R15

The day the same football did not work

1-2H · Suwon Samsung

The highest identity score of all 23 matches: 37% possession, 314 passes, average pass 21.9m — on the numbers, the most Hwaseong-like game of the season. And yet only one offside, from a side that averages 2.4. The runs behind were not even being caught, which points to the league leaders dropping their line and erasing the space the whole model attacks. The result: eight shots, 0.89 xG. The only goal came from a corner — open play did not break them.

Watch for — Does Hwaseong have a plan B against a side that erases the space behind? This match is the model's ceiling.

37%Possession

1Offsides (avg 2.4)

0.89xG

2.98xG against

The control — R7, home to Jeonnam, 1-0

The lowest identity score of the season (−1.21): 64% possession (season high), 664 passes (season high), directness 17% (season low). They won, but off 15 shots worth 1.48 xG while conceding twelve. Evidence that they can hold the ball if asked — and evidence that they become a different side when they do. Set beside the four above, it sharpens what the default actually is.

R	H/A	Opponent	Res	Poss	Passes	Direct%	Shots	xG	OFS	Index
15	H	Suwon SB	L 1-2	37%	314	67	8	0.89	1	+0.68
20	H	Daegu	W 5-1	35%	268	70	13	1.99	2	+0.61
24	H	CB Cheongju	W 2-1	35%	327	83	13	1.50	4	+0.60
12	H	Busan	W 3-2	46%	436	75	16	2.63	3	+0.38
8	H	Gimpo	D 2-2	43%	343	40	13	1.69	3	+0.32
1	A	Daegu	L 0-1	42%	369	44	11	1.62	2	+0.31
11	H	Suwon FC	D 1-1	38%	332	75	11	1.54	2	+0.30
13	A	CB Cheongju	W 3-2	52%	466	57	8	0.90	6	+0.24
10	H	Ansan	W 2-0	43%	453	43	15	2.07	3	+0.09
19	H	CN Asan	W 3-2	41%	336	33	11	1.64	1	+0.00
21	H	Seoul E	D 0-0	41%	453	0	8	0.75	5	-0.01
14	H	Gyeongnam	W 2-0	47%	444	33	15	2.15	3	-0.02
22	A	Busan	D 0-0	44%	414	23	21	2.20	5	-0.05
18	A	Yongin	W 3-0	42%	317	29	9	1.63	0	-0.07
9	A	Seoul E	W 2-1	40%	385	50	6	1.40	3	-0.10
3	H	Cheonan	D 2-2	45%	412	50	18	1.67	0	-0.13
6	H	Seongnam	L 1-3	46%	433	56	10	0.73	1	-0.17
2	H	Gimhae	W 2-0	58%	579	36	11	1.73	2	-0.19
23	A	Jeonnam	L 0-1	47%	408	33	7	0.94	0	-0.26
17	H	Paju	L 1-2	59%	523	36	16	1.15	3	-0.27
4	H	Yongin	D 0-0	49%	531	33	6	0.40	2	-0.52
5	A	CN Asan	L 0-1	56%	538	43	7	0.64	5	-0.52
7	H	Jeonnam	W 1-0	64%	664	17	15	1.48	0	-1.21

All 23 matches, sorted by identity index. The four highlighted rows are the ones chosen above. The 0% directness against Seoul E in R21 means every shot-producing possession that day took four passes or more.

11So what should be watched

If you are facing them

The space behind the last line. League-most offsides (2.43) is a proxy for how often they run. When that fell to one — R15 against Suwon — their xG was 0.89.
The channels. 83% of their entries into the final third are wide. Winning the flank duel and blocking the cross matters more than sealing the half-space.
The 46th to 60th minute. Ten scored, one conceded in that quarter-hour. Survive the restart and the game looks different.
The opening fifteen. Their weakest window — opponents have already taken 40 shots there against Hwaseong's 26.

If you are Hwaseong

Eight goals conceded after the 76th minute. Cost of the style or a management problem — the two cannot be separated without physical data.
50% reliance on Plana. Involved in 18 of 36 goals. No alternative shows up in the data if he is missing.
5.3 corners a match, three goals. The league's third-highest supply, almost unconverted. The cheapest improvement available.
No first-half goal in 12 of 23. Still unresolved whether the second half is well designed or the first half is simply empty.

Appendix — data, verification, limits

Source

Everything comes from the Korea Professional Football League's K League Data Portal (data.kleague.com), public data accessible without a login. No figure here was read off a screenshot; the responses the portal itself uses to draw its pages were parsed directly. Collected 2026-09-04, covering the 23 matches Hwaseong have played through round 24. K League 2 has 17 clubs, so one rests each round — Hwaseong's byes are rounds 16 and 33.

Cross-checks

League table vs re-summed match records: exact match (23 games, 11-6-6, 39 points, 36-24)
Sum of per-player match data vs team season totals: exact match (271 shots, 9,239 passes, 488 crosses, 202 key passes, 570 tackles, 212 interceptions)
Corners and offsides from match records vs season totals: exact match (121 / 56)
Opponent goals in the event feed vs official goals against: exact match (24)

Data issues found

The 'total' cell of the goals row is wrong in eight of the portal's match records. Summing those cells gives 32-17; re-adding first and second halves gives 36-24, which is what the official table says. Every scoreline here uses the re-added figure. Every other row — shots, corners, fouls, offsides — agreed across all 23 matches.
"How many shots Hwaseong took" depends on the source: 271 in the supplementary records, 268 in the event feed, 264 in the portal's xG table, 240 in the match records. The portal runs more than one counting system. League comparisons here use 271, coordinate and time-band analysis uses 268, and the two are never mixed.
The event feed tags 34 of the 36 goals (94%). One own goal cannot appear as a Hwaseong event by construction and one is missing. The match record is the authority for goals; the event feed is used only for spatial and time-band work. xG differs too — 33.35 summed from events against the portal's official 30.22. Team-level xG here uses the official value.

My own definitions (not official metrics)

Possession phase — one side's spell until the opponent registers two consecutive events or the keeper collects
Directness — share of shot-producing phases containing three or fewer passes by that side
Wide entry — of passes starting below X 66.7 and ending at or beyond it, the share landing in the outer thirds of Y
Identity index — the mean z-score of the eight axes listed in §10

Coordinates

The portal normalises X and Y to 0-100 but keeps them fixed to the pitch, so attacking direction flips at half-time. Every positional figure here was computed after rotating each team so it always attacks towards increasing X. Check: after rotation the mean shot origin sits at X 88.0 for Hwaseong and 86.4 for their opponents — both correct.

What this piece does not answer

What actually changes after the break — instruction, opponent fatigue, or substitution. The substitution timing (70th minute on average) does not explain it.
Who triggers the runs in behind, and when. Offside counts are only a proxy for frequency.
Whether avoiding duels is a design choice or a consequence of the squad's physical profile. Event data cannot tell the two apart.

All three need video and inside information. Marking where the data stops is the rule of this series.