Near Post or Far Post: The 90+6 Equation Football Models Cannot Measure

The 90th minute plus six. Score 0-0. The referee had the whistle to his lips. A direct free kick at the edge of the box, in the truest sense the last chance of the match. Hiroaki Okuno, 37 years old, placed the ball, took three steps back, and chose the far post. Not the near post. Not a cross into the five-and-a-half-metre box for a centre-back to head home. The far post — the narrowest window on the goal, where the ball must clear the wall, bend beyond the goalkeeper's reach, then drop into the top corner. The ball went in. Shonan Bellmare won. Those six seconds produced a result. Kosuke Ota — 348 J.League appearances and seven Japan caps — saw a tactical lesson inside it. Ota is not a man who speaks just to fill air. A left-footed left-back, he is known across Japanese football as a set-piece specialist of rare precision. After retiring he became a club ambassador for Machida and runs free soccer schools nationwide. When a man with that technical CV speaks about a free kick, I listen — because he is not speaking in emotion, he is speaking in mechanism. In the post-match broadcast, Ota used a telling phrase: "everything clicked together." He described the ball travelling with speed, curve and drop all at an ideal level. But the crux of his point sat elsewhere, and it is the reason I rewound the tape: if it were him, Ota said, he would have chosen the near post. The near-post logic is simple, and it is the default logic of attacking football. Strike the near post and the goalkeeper has less time to react. And you always have an escape route: if the ball does not go in, it can spill out for a teammate to charge onto. A near-post strike a few centimetres off target still carries value. An overhit near-post strike still carries value. It tolerates error well. The far post is the inverse. A smaller window, a longer path for the ball, and if the goalkeeper reads the direction early, he has enough time to reach it. As Ota put it, far-post attempts miss by more than people imagine. It is a gamble. But this is the point where I — someone who has spent hundreds of hours logging dead-ball situations by hand — want to separate the mechanism from the emotion. Ota's near-post logic is the logic of a game still in progress. It assumes that after the strike, the match continues. There is another minute, another phase, another chance for a teammate to pounce on a rebound. In an ordinary game state that assumption always holds, which is why the near post is the probability-optimal choice. At 90+6, that assumption vanishes. There is no phase after the strike. Nobody is charging in for a rebound, because if the ball spills out the referee will have blown full time before it touches the ground a second time. The value of the "escape route" — the only thing that justifies the near post — falls to zero. When that value reaches zero, the whole equation changes. You are no longer optimising the probability of scoring across a sequence of actions. You are optimising the probability of scoring across one single action, the last one, with no correction possible. In that equation, the far post stops being a gamble — it becomes the rational choice, because the near post has lost its greatest reward. This is the kind of decision that match-tracking models cannot measure, and I say that as someone who builds them. xG has no variable for "seconds remaining." PPDA has no variable for "who is charging in for the rebound." My model does not cry and does not celebrate, but after every match it owes me a lesson — and the lesson here is that game state is a variable, not a constant. Mechanically, Okuno's strike followed the textbook. Ota described him keeping a clean contact on the ball, letting the standing leg load it rather than driving through it. That contact produces three things at once: speed, curve and late drop. It is the signature of an inswinger — the ball travels outward, then bends back in. Normally a free-kick taker must trade speed against curve; Okuno had both, plus a drop that stopped the goalkeeper changing the direction of his hands in time. The technique is not new. What is new is choosing it in this specific situation. Ota pushed the story to another layer: the psychological duel with the goalkeeper. He said Okuno won the mental battle. The most reasonable reading is that the goalkeeper stood shaded toward the near post, the standard reflex of a man who has read thousands of free kicks. Once the goalkeeper leans near post, the high corner on the far side becomes the only genuinely scoreable window. Ota called it "the only possible spot." I mark that claim at medium confidence, because it is an inference from a qualitative quote, not from goalkeeper-positioning data. There is one more variable I always check: age. A 37-year-old loses running and duelling output — physical data shows no mercy. But dead-ball technique is the slowest-declining asset a player owns. The fact that Shonan handed the decisive final free kick to a 37-year-old rather than a younger teammate is a signal about the hierarchy inside the squad. It is also a signal about the ranking of skill: when a match is decided by a static moment, experience remains the strongest currency. Based on my experience tracking matches, late-game free kicks in Japan tend to follow a fairly fixed template: either a cross into the box for a centre-back to head, or a hard strike straight at the near post. Choosing the far post here broke the template, and that very deviation is what caught the goalkeeper out. This is where data and instinct touch. What I want to question is the tail of the story, as it spreads from one moment into a run of results. After the match, the commentary appeared: Shonan have momentum, morale is rising, the team is hitting form. These lines sound sensible, and they match the feeling. They do not match the denominator. A free-kick goal at 90+6 creates a maximum swing of one point in the table compared with a draw. One point. Over a 38-round season, taking a single moment and inferring a trend from it is methodologically wrong. With the data I have — fourth place, 11 points from six matches — I can only say Shonan are tracking expectation, slightly above the baseline. I cannot say more, and I decline to say more. Here I have to state plainly a technical problem I found when cross-checking sources. The report labels this match as J2 League, round six, but the fixture is Shonan Bellmare against Vanraure Hachinohe. Under the structure of Japan's professional league, Shonan play in J1 and Vanraure Hachinohe play in J3. These two clubs do not meet in a regular J2 round. On top of that, a sixth round usually falls in early spring, not on 12 September. I am not concluding the source is wrong — I am saying that before using any table number from that report to project the future, a reader should check it against the official fixture list. Protecting the integrity of numbers means naming the gaps in the source you are using. One more thing about models. I do not trust managers, I trust models. But I listen to managers in order to fix models. In this case, I listened to Ota. His reading of near post versus far post is a reusable decision framework, and it reminded me that every set-piece model needs one variable I had never entered: the state of the clock. Okuno's free kick will fade within weeks, but the question it leaves behind will not. When should a player choose the far post? Only in the final minute, or also in other situations where rebound value has already disappeared? I will open the model, add a variable for time remaining, and let it answer. The first xG table I wrote by hand on a bus, back when nobody called it data. Now it still owes me one more lesson. The crowd watches the ball; I watch 22 numbers moving — and wait patiently for them to tell a different story.