Three Corrections Before a Defensive Number Means Anything
Words & data analysis | Choi Bong-jin (Far Post Analytics operator)
Two goalkeepers top the 2024 J1 duel win-rate table. Tackle leaders play for teams that concede possession. Three corrections to apply before reading defensive metrics, with J1 data.
Data: API-Football · 2024 J1 League season
Key takeaways
- Tackle and interception counts measure defensive opportunity as much as defensive ability. A team that concedes possession manufactures tackles for its players.
- Goalkeepers register duel wins when claiming high balls. In our 2024 J1 sample they occupy the top two duel win-rate positions at 89.3% and 78.6%, above every outfielder.
- Restricted to outfielders, the leaders are centre-backs — 65.2%, 64.8%, 62.1% — a range roughly 15 to 25 points below the contaminated figures.
- The hardest correction is that elite positional defending produces ordinary counting stats. A player who prevents the situation never records the event.
Sort the 2024 J1 duel win-rate table with no conditions and the top two entries are goalkeepers. That is not a quirk to laugh at and move past — it is the clearest available demonstration that defensive metrics do not survive naive reading.
Why attacking and defensive stats are not symmetrical
Attacking metrics have a convenient property: the direction of goodness is stable. More goals is better. More key passes is better. The number can still mislead about volume versus efficiency, but it does not reverse.
Defensive counting stats have no such property. A high tackle count is consistent with a player who defends well and with a player whose team spends the match retreating. Those are opposite readings from an identical number, and nothing in the column distinguishes them.
Three corrections handle most of the damage. None of them is optional.
Correction 1 — tackle counts measure opportunity
In our 2024 J1 detailed-stats sample the tackles-per-90 leaders are Daiki Kaneko of Júbilo Iwata at 2.64 and Daihachi Okamura of Machida Zelvia at 2.60. The figures are near-identical. The circumstances producing them are not.
A player at a side under sustained pressure faces more defensive situations per match. More situations means more attempted tackles, whether or not the player is any good at them. At a possession-dominant club the same defender’s count contracts, because the opportunities do not arrive.
This makes tackles and interceptions unreadable without team context. A defender registering 1.5 tackles per 90 for a side that controls the ball may be doing more difficult work than one registering 2.5 for a team defending its box every week. The direction of the adjustment is clear; the size of it is not, which is why we treat this as a discount applied by judgement rather than a computed factor.
The practical rule: never read a tackle or interception rate without simultaneously knowing the team’s possession share and league position. If those are unavailable, the number is not usable.
Correction 2 — cohorts must exclude goalkeepers, and then split by position
The duel win-rate table in our 2024 J1 sample is led by Vissel Kobe's Daiya Maekawa at 89.3% and Júbilo Iwata's Eiji Kawashima at 78.6%. Both are goalkeepers.
The mechanism is mundane: a keeper claiming a cross or high ball is frequently logged as a duel won, and he contests almost nothing else. His sample consists disproportionately of the one duel type he is structurally favoured to win. Leave keepers in an outfield ranking and the top of the table is simply wrong.
Remove them and the picture is coherent.
| Player | Club | Pos | Win % |
|---|---|---|---|
| Daiya Maekawa | Vissel Kobe | GK | 89.3% |
| Eiji Kawashima | Júbilo Iwata | GK | 78.6% |
| Daihachi Okamura | Machida Zelvia | DF | 65.2% |
| Kawazura | Nagoya Grampus | DF | 64.8% |
| Nakano | Sanfrecce Hiroshima | DF | 62.1% |
Removing goalkeepers is the first step, not the last. Within the outfield group, duel win rate still splits sharply by position — defenders contest structurally winnable battles, forwards do not. A 55% figure is unremarkable for a centre-back and strong for a striker.
This is why every percentile on this site is computed against a position-specific cohort. A leaderboard that mixes positions is reporting position under a performance heading.
Correction 3 — the best defending produces no events
The third correction is the one no adjustment fixes, because the problem is not distortion but absence.
A defender whose positioning closes a passing lane records nothing. The pass is not attempted, so there is no interception. A centre-back who reads a run early and arrives before the contest is necessary records no tackle and no duel. The event that would have generated the data point was prevented from occurring.
This produces a specific and expensive error: a screen built on counting stats systematically favours high-activity defenders over positionally dominant ones. Both categories exist and both are valuable, but they are not the same signing, and a department that only reads tackle and interception volume will keep finding the first while believing it has searched for the second.
No metric in our dataset resolves this. Ordinary counting numbers should never be treated as disqualifying on their own — and we would rather state that limitation than imply the database sees something it does not.
What the three corrections do and do not fix
Automated in our database: position-specific cohorts, goalkeeper exclusion from outfield rankings, minimum-sample floors, per-90 normalisation.
Applied by judgement, not computed: the team-context discount for possession share and league position. We do not publish an adjustment factor because we do not have a defensible one.
Not addressed by data at all: the absence-of-events problem. This requires footage and live viewing, and there is no version of our screening layer that substitutes for it.
Stated plainly: the screen narrows the list of grounds worth travelling to. It does not rank defenders, and it should not be used as though it does.
Caveats and limitations
- Ground and aerial duels are not separated in this dataset, which limits how far the outfield rankings can be interpreted.
- Tackle and interception success rates are not distinguished from attempts in the figures cited above.
- The team-context correction is applied by judgement. No published adjustment factor supports it.
- Figures are drawn from our detailed-stats sample rather than the full 2024 J1 league. Players outside that collection are absent, not ranked low.
- Position labels are provider-assigned and may not match the role actually played.
- Single season, 2024. No multi-season stability check has been applied.
Frequently asked questions
Does a high tackle count mean a player is a good defender?
Not by itself. Tackle volume rises with the number of defensive situations a player faces, which is largely determined by how much possession his team concedes. A defender at a side under regular pressure will accumulate tackles regardless of quality, while an equally capable defender at a possession-dominant club records fewer. Read the rate alongside team possession share and league position, or not at all.
Why do goalkeepers appear at the top of duel win-rate tables?
Because claiming a cross or high ball is commonly logged as a duel won, and a goalkeeper contests little else. His sample is concentrated in the one duel type he is structurally favoured to win. In our 2024 J1 sample this put two keepers above every outfield player, at 89.3% and 78.6%. Any outfield duel ranking must exclude goalkeepers before it is read, and then split the remaining players by position.
Can data identify a positionally excellent defender?
Not from counting statistics, and we do not claim otherwise. A defender who closes a passing lane through positioning prevents the pass from being attempted, so no interception is recorded; one who reads a run early arrives before a tackle is required. The best defensive work removes the very events that would have produced the data. Counting stats reliably identify high-activity defenders; identifying dominant ones remains a task for footage and live observation.
Sources and disclosure
- Data: API-Football — 2024 J1 League season. Tackle, duel and position figures as retrieved for the Far Post Analytics database, restricted to players with detailed stats collected.
- Calculation: Per-90 conversions and duel win rates are Far Post Analytics calculations. Percentiles are computed within position-specific cohorts. See the Methodology page.
- Known data limitation: Ground and aerial duels are not distinguished in the source feed. Preventive defensive actions are not recorded by any metric in this dataset.
- Data vintage: 2024 J1 season. Not updated with subsequent seasons.
- Disclosure: Far Post Analytics is an independent analysis publication. It does not represent players, does not broker transfers, and receives no compensation from any club or agency. See our Editorial Policy.
Sources last verified August 1, 2026
Data sources and season coverage are stated within each article; where a report carries its own sources section, that section governs. Ages are as of data collection. Per-90 metrics are our own calculations, and the smaller a player's minutes sample, the wider the margin of error. Every number here is a starting point for scouting — never a substitute for it.
✍️ Choi Bong-jin
Operator of Far Post Analytics. I analyze scouting data for the J.League and Asian football. My goal is to find the next transfer-market star where Europe isn't looking.
About the operator →