Duel Win Rate Measures Position First: Reading the 2024 J1 Cohort Correctly

Thirteen of the top twenty duel winners in 2024 J1 are defenders. None are attackers. That is not a finding about which players are strong — it is evidence that an unsegmented duel leaderboard is a position chart wearing a performance label.

A scouting note converted into the wrong number

"Strong in a physical battle" is one of the most common lines in a scouting report, and duel win rate is the metric it usually becomes: duels won divided by duels contested, ground and air combined. The translation looks clean. It is not.

The problem is that duels are not a uniform event type. A centre-back stepping across an attacker running into traffic in his own half is engaged in a structurally favourable contest. A striker with his back to goal, holding the ball against a defender who has the angle, is in a structurally unfavourable one. Both are logged identically.

Rank a league on that column with no conditions and the output is largely determined before any player steps on the pitch.

The size of the positional effect

Splitting the same 74-player cohort by position gives the scale of the distortion directly.

The gap between defenders and attackers is 13.9 percentage points. For context, the gap between the best defender in the league and the defensive average is roughly 14.4 points — meaning the positional effect is almost exactly as large as the entire span from average to elite within a position.

Any ranking that does not remove that effect first is reporting it as if it were performance.

What the correction does to individual readings

Subtracting the positional baseline changes the interpretation of specific figures, sometimes reversing it.

A forward at 50% sits below the cohort average and looks unremarkable on a league table. Against a 42.2% attacker baseline he is roughly 8 points clear — the equivalent, in positional terms, of a defender at 64%.

A centre-back at 54% is above the cohort average, comfortably mid-table on a league list, and roughly 2 points below the defensive baseline. For a first-choice centre-back that is a flag, not a neutral reading.

A midfielder at 51% is almost exactly average both ways, because the midfield baseline sits within a point of the cohort figure. This is the one position group where the uncorrected number is roughly usable — which is precisely why it misleads people into trusting the metric generally.

Only the third case survives an unsegmented reading. The first two are actively wrong, and the first is the more costly error for a recruitment department: it discards a player who is performing well in the hardest duel environment on the pitch.

Two details inside the leaderboard

Age is not the driver it is assumed to be. Shiro Sasaki was 36 in this dataset and ranks sixth at 64.9%, on a substantial 336-duel sample. Duel success is decided by reading the situation early, body position at contact, and timing — attributes that do not decline on the same curve as sprint speed. A duel table is a poor place to look for an age cliff.

The young defenders are the scarce commodity. The youngest players in the leading group are 25: Kosuke Mikuni and Ikuma Sekigawa. Everyone else is 26 or older. For a recruitment department working an under-25 brief, that scarcity is the actual finding here — not the ranking itself but the shape of its age distribution.

Neither observation should be read as a transfer recommendation. Both are reasons to look, and this data alone does not support anything stronger.

What duel win rate cannot tell you

Correcting for position fixes the largest distortion. Several remain, and a method that reports the correction without the residual limitations is only half-honest.

1. Ground and aerial duels are combined. A tall centre-back who dominates in the air and a quick full-back who wins on the ground produce the same column value. For most recruitment questions these are different players.

2. Exposure is not measured. The rate says nothing about how often a player is put into duels, or where on the pitch. A defender in a deep block contests different battles from one in a high line, and volume is a team property as much as a personal one.

3. A won duel is not a good outcome. Winning the ball and immediately conceding possession is logged as a win. The metric records contact, not what followed it.

4. Position labels are provider-assigned. The DF/MF/FW split above is only as good as the classification, and a player whose recorded position does not match his actual role will be measured against the wrong baseline.

This is the reason our player map and percentile displays rank within position groups by default. It does not make duel win rate a strong metric. It makes it a readable one.