Suwon Samsung · Tactical proposal

You do not need more crosses

The club that crosses more than anyone in the league ranks 6th for goals. Setting 21 matches of the 2026 K League 2 season against 96 Big 5 clubs and the 2025 K League technical report narrows the leak to a single phase — the point where a cross becomes a shot. What needs changing is not the volume but the conditions in front of it. Twelve figures for the eight-act proposal, each paired with why this figure — what a first-time viewer reads in three seconds.

2026 K League 2 · 21 of rounds 1–22 (no R12 data)12 figuresDraft v1
ACT 0 · HOOK

Same players, a different cross

Suwon's corners are a weapon the league already acknowledges. The open-play crosses those same players put into that same box are the most numerous in the league — and the least likely to become a shot.

Why this figure Two tactical boards need no explanation. On the left there is no presser and five bodies stand in the box; on the right there are two pressers and two bodies. What the two pictures share is the players; what differs is only the conditions. It closes off the "we lack the personnel" objection in the opening scene.

① Same box, two scenes — corner vs open-play cross

Final 40m of the attacking half · goal at the top

Suwon Opponent dashed arrow = cross path
A corner is delivered, by definition, with zero pressure, unlimited time and the box already set — 110 of them this season, 2nd in the league, and the federation's technical report named Suwon as its K League 2 set-piece case. The right-hand board redraws a scene that recurred in the R7 Gimpo match, following the qualitative notes (positions are illustrative, not measured coordinates). Open-play crosses came 554 times, 1st in the league, and converted to shots 49.1% of the time — the Big 5 figure is 71.6%.
"When there is no pressure, these players cross very well. There are good kickers here and the corners are all sharp." R10 Suwon FC (A) · qualitative notes
ACT 1 · PROBLEM

The most produced, the least collected

Why this figure A single shape — a staircase collapsing from top to bottom — finishes the problem statement for the whole proposal. Not "Suwon cannot produce", but "production leads the league and disappears at collection." The z-score needs one sentence: league average is 0, one step is one standard deviation.

② Production · output · efficiency, against the league average (z-score)

2026 K League 2, 17 clubs · season totals

Production — how much was made Output · efficiency — how much was collected
Blue = how much was made; orange = how much was collected. Every production metric sits at +1.0σ or above, goals at +0.22σ, and goals per shot at −0.45σ — below average.
Shorthand in this figure z-score a ruler for how far a value sits from the league average. 0 = the league average, and one step (1σ) is one unit of how widely the 17 clubs spread around it. +2σ means effectively the highest in the league; a negative value is below average.
One sentence The club that makes more crosses, more final-third passes and more key passes than anyone in the league ranks 6th for goals.
ACT 2 · LOCATE

There is only one leak

Why this figure It answers "where is the problem?" with a picture in which only the first of three gates is low and the other two match Europe. It does not name a culprit — it narrows the search. Finishing and composure are ruled out here.

③ Pass rate by phase — three gates from cross to goal

96 Big 5 clubs · 12 K League 1 clubs · 17 K League 2 clubs · Suwon over 21 matches

Big 5 leagues K League 2 K League 1 Suwon Samsung
The gap opens by 22.5pp at the first gate alone; across the next two, Suwon overlaps Europe to one decimal place. This is not a side that fails to score. It is a side whose crosses do not become shots.
Shorthand in this figure Pass rate how many of 100 at one stage reach the next. 100 crosses → 49.1 shots is 49.1%. pp percentage points — the gap between two percentages. 71.6% − 49.1% = 22.5pp, i.e. 22.5 notches apart on the percentage scale, not "22.5% different".
Why this figure If the funnel shows Suwon's internal structure, the scatter shows where Suwon sits on a map of 96 clubs. One dot is one club and only Suwon is orange. Being furthest right reads without a caption.

④ Crosses per shot vs goals per match — a map of 96 Big 5 clubs

2025/26 Premier League · LaLiga · Serie A · Bundesliga · Ligue 1

The dashed line is the trend across 96 clubs (r = −0.68). The worse the cross-to-shot conversion, the fewer the goals. Suwon sits further right than any of the 96. Inter Milan put in 23 crosses a match and still sit top-left — the built-in rebuttal that this is a conversion problem, not a volume one.
Shorthand in this figure r how closely two values move together. 0 is unrelated; ±1 is locked together. Here r = −0.68 means clubs that spend more crosses per shot clearly tend to score less. A tendency, not a law.
ACT 3 · CAUSE

The cross may be a result, not a choice

Why this figure The staircase only goes one way. Monotonic ordering with no exception is the strongest picture available to a reader who does not read statistics. "The more the ball went through midfield, the more chances followed" ends in four bars.

⑤ Output by quartile of the central-routing index

21 matches split into quartiles by routing index · 5 matches each

Central-routing index = entry rate into midfield × forward-progression rate out of midfield, computed by parsing both teams' pass matrices across 21 matches. 2.1× on xG, 1.8× on shots, monotonic across all four bins. Caution — this is an observed association, not proof of cause. Five matches per bin.
Shorthand in this figure Q1 · Q2 · Q3 · Q4 the 21 matches lined up by routing index and cut into four groups (quartiles). Q1 is the five matches with the least routing, Q4 the five with the most. These are group averages, not individual matches.
xG expected goals — how many goals those shots become on average, converted from shot location and situation. xG 1.96 reads as "a match producing chances like this usually yields about two goals." Actual goals differing from it is normal — that difference is finishing.
And this is the one area where Suwon matches the opponent Comparing both teams' pass matrices under the same rule across 21 matches gives Suwon 5.95% against opponents' 5.51% — statistically indistinguishable (t = 0.44). The club that makes more crosses, final-third passes and key passes than anyone in the league is level with its opponents at routing the ball through the centre of midfield. It does track results monotonically, though — wins +2.43pp · draws −0.45pp · defeats −3.70pp. Caution — four defeats is a very small sample. Read the direction, not the size.
What t means t is a ruler for whether a difference is a real one or just the range a figure swings through from match to match. It is the size of the difference measured against that match-to-match spread, and roughly 2 is where one starts calling it a difference. Here t = 0.44 means the gap (+0.44pp) is less than half the match-to-match spread — Suwon and its opponents cannot be told apart on this measure.
ACT 4 · MECHANISM

Why that cross does not turn into value

Why this figure The whole point is that these multiply rather than add. A ribbon leaves the left edge and narrows through four gates, and where one gate halves it, half of it never comes back however good the rest are. Laying two rows over each other makes which gate separated them something you can point at.

⑥ Value of a cross = Z × F × A × Del

Two matches a week apart, passing through the four gates

Only the two ends are recorded fact — R7: 29 crosses → 4 shots (13.8%); R8: 35 crosses → 13 shots (37.1%). The four intermediate drops distribute the qualitative grades as an illustration and are not measurements. R7 lost roughly half at A (arrival) and half again at Del (delivery) — 1 of 29 landed in the target band — while R8 recovered A by switching to two strikers and tripled its shots from the same volume of crosses. T (time and context) is kept as a recorded condition, not promoted to a metric.
ACT 5 · CONTROL

It is the conditions, not the ability

Why this figure The hook thrown in act 0 is paid off here. Two matches a week apart, and two matches whose opponent and output volume coincide — pairs where the players, the coach and the training were the same and the result diverged. A controlled comparison is the cheapest way to close off "we were unlucky".

⑦ R7 ↔ R8 — consecutive rounds, seven days apart

Last 4-3-3 (0-1 L) ↔ first 4-4-2 (1-0 W)

R7 GimpoR8 Gyeongnam
Crosses barely moved (29 → 35) while shots in the box went 2 → 9. What changed was not the volume delivered but the number of bodies arriving in the box (A). Caution — one pair of matches. An illustration of direction, not a test.

⑧ R10 ↔ R22 — same opponent, output matched to the decimal

Suwon FC away (1-3 L) ↔ Suwon FC home (2-2 D)

R10R22

↓ and yet one thing flipped

How the value is derived

Central-routing index = entry rate × progression rate   ·   Routing edge = Suwon's index − opponent's index
EntryProgressionIndex
R10 Suwon27.1%8.0%2.17← lowest of 21
R10 Suwon FC31.3%12.7%3.97
−1.79edge · 17th of 21
R22 Suwon36.9%15.0%5.53
R22 Suwon FC40.3%5.6%2.26
+3.27edge · 5th of 21
The two gates multiply because the ball has to clear both to count as "routed" — reaching midfield without going forward is not routing, and progressing well without the ball arriving there is not routing either. Look at R22 Suwon FC: entry is high at 40.3% but progression is 5.6%. That is the same scene as the qualitative note, "Suwon FC played a lot of long balls, so fewer threatening moments came". Both teams were computed under an identical rule (portal position labels, substitutes excluded). This is the best-controlled single piece of evidence in the project — the volume produced was the same, and only the edge in midfield flipped.
"In the out-channelling, one central midfielder pushes wide and the other drops to cover the centre-back. That leaves the centre empty underneath the two strikers. … In K League 1 the leakage and exposure there would have been severe." R22 Suwon FC (H) · qualitative notes — the data reached the same conclusion afterwards
ACT 6 · PRESCRIPTION

What can be done

Why this figure This is the page people turn to first, which is why it is cards rather than a chart. Every card carries evidence · form · when to start in the same position so they can be compared. Top-left is what can be done soonest; further right takes longer.
FIRST · START NOW

Delivery band and agreed runs

Inswing band drills — the flanks of the six-yard box, behind the defenders. The crosser's target and the runner's route are agreed in advance.

It changes neither the personnel nor the structure. Fastest to run and fastest to show.

R7 · 1 of 29 in bandDel · Fit-R
SECOND · IN SEASON

Producing transitions

Repeating the cross situation inside three seconds of a large switch. Lure and Shock moved onto the time axis as a drill.

T is not promoted to a metric. It stays a recorded, observed condition.

transition → accuracy r ≈ +0.4T
THIRD · BEFORE PROMOTION

Progression through midfield

Lateral-pass protocol · chained post play. Build a route that passes through the centre once before being pushed wide.

It will not be finished inside a season. That is why it is the one item to start now.

Zone 8 · 2.94×quartile xG 2.1×
PRECONDITION

Balance in both phases

Cover when the third line is beaten. This is the precondition for track 2 (more bodies in the box) and the order is not reversed. With the back exposed, no mutation can be attempted.

2026 data only.

7 observationstrack 2 premise
What does not need doing More crosses (1st in the league) · more box entries (2nd) · cross-accuracy work (completion 5th) · running more (workload is identical across the two leagues, and Suwon is already below average). The proposal names what to change, not what to add.

The blue-underlined terms — Lure and Shock · lateral-pass protocol · chained post play · pre-agreed pattern · variable and mutation — are the concepts this proposal is built on. Each opens in a new tab on the page that explains it with an animation.

ACT 7 · WHY NOW

After promotion, the improvement becomes a condition of survival

Why this figure The size of each gap sits on a real pitch as a circle. What is eighteen rows in a table becomes, on a pitch, one reading of "where is the gap". One large circle in the centre of midfield, and only a dot in the opponent's box — that is the whole figure.

⑨ K League 1 ÷ K League 2, VAEP ratio by zone

2025 K League technical report · circle size = size of the gap

Where K League 1 outperforms K League 2 is not the opponent's box. 74% of the gap arises in the defensive half and midfield, and within that, midfield centre dominates at 2.94×. Suwon's difficulty using the centre is not Suwon's alone but a property of K League 2 as a whole — which is exactly why it is the first thing a promoted side runs into.
Shorthand in this figure VAEP a single value summing how much an action (a pass, a carry, a defensive act) raised the chance of scoring and lowered the chance of conceding — "that was a good play", put into a number. Only the ratio between the two leagues in the same zone is used here: 1.00× means the leagues are level, and 2.94× means K League 1 generates nearly three times the value in that zone.
Zone 1–18 the pitch cut into 6 columns × 3 rows. Our goal is on the left, the opponent's on the right.
Why this figure A figure that corrects the common line "K League 2 is simply weaker" to half true. The right ends of the two lines touch and only the left ends separate → K League 1 is not a league of stronger opponents; it is a league with no easy fixtures.

⑩ Top-4 and bottom-4 averages, two leagues compared

2025 K League technical report · total team VAEP

The top is level at −2.3%; the bottom separates by −17.2%. Coefficient of variation: 9.6% in K League 1 against 15.8% in K League 2. In K League 2 a good part of the season is made up of winnable fixtures. In K League 1 there are none of those. 38 matches against 32 — six more of them.
Shorthand in this figure Coefficient of variation how widely the clubs are spread, expressed relative to the mean. Lower means the clubs are more evenly matched. 9.6% in K League 1 against 15.8% in K League 2 — the second tier is spread 1.6× as wide.
ACT 8 · HONESTY

What we got wrong

Why this figure A list with lines struck through it builds trust in the method by itself. The last three rows are not hypotheses about Suwon but the proposer's own unverified theory. A list that only refutes other people's claims and a list that puts its author's on the same page land very differently.

⑪ Refutation log

Hypotheses rejected or corrected during verification

HypothesisVerdictEvidence
Cross completion is the problemrejected24.4% · 5th in league (z = +0.67)
Cross volume degrades shot qualityunsupportedmatch level r = +0.097
The switch to 4-4-2 failedrejectedxG per match 1.01 → 1.75 (+73%)
Shot quality fell after the switchrejected−27% on goals, but xG/shot 0.104 → 0.109 unchanged — it was finishing variance
Ilyuchenko's transfer broke the attackrejectedafter the transfer, shots 13.61 → 13.67, shots in box 8.8 → 9.0
4-4-2 degraded the defencerejectedR8–19 opponent shots 7.91, in box 5.00 — unchanged throughout
Opponents target Suwon physicallyunsupportedall six metrics t < 1.5 · adjusted for possession, Suwon engages 30% more often
K League 2 plays more matches than K League 1rejectedK1 38 matches / K2 32
Pass-length diversity = tempo variationproxy failedwhat it measured was "how far the side was forced long"
K League 2 is weaker across the boardhalf truetop 4 −2.3% / bottom 4 −17.2%
Proposer's theory — L3 (threshold of repeated exposure)unverifiedK ≈ 4 by appearance, but a test needs 48 matches
Proposer's theory — bufferingunverifiedtriangulated at best; not a proof
Proposer's theory — L2unobservablecannot be observed at all with the data currently available
The three orange rows are the proposer's own theories, not Suwon's hypotheses. And one methodological rule remains — never judge efficiency from goal-based figures alone; always cross-check against xG. This project fell into that trap twice.