A pressing trigger is labelled after we already know what happened

Coding triggers from video means the successful presses get called triggers and the identical failures get called nothing.

We decide it was a trigger because the press worked.

That is the circularity at the heart of most published pressing analysis, and once you notice it you cannot stop seeing it. An analyst watches the match back and marks the moments the press was activated. A backwards pass, a heavy touch, a full back receiving with his body shape wrong. Those are the trigger events. Except the analyst is watching with the outcome already visible, and human coders, however conscientious, are more likely to mark a moment as a trigger when the sequence that followed looked coordinated.

The identical cue, ignored or mistimed, producing nothing, tends not to get coded at all. So the dataset contains successful triggers and almost no failed ones, and any subsequent analysis of trigger effectiveness is computed on a sample selected on the outcome. You will conclude that triggers work. The sample was assembled by keeping the ones that did.

The correct design is straightforward in principle and tedious in practice. Define the cue physically and in advance: opponent receives facing his own goal within a defined zone, ball travel time above some threshold, first touch that moves the ball more than a set distance. Then detect every instance automatically, whatever followed, and only afterwards look at what the pressing side did. Now you have a denominator. Now you can ask how often the team actually pressed when the cue occurred, which is the question coaches genuinely want answered, and it cannot be answered from video coded backwards.

There is a related error in how trigger discipline is discussed. Teams are said to press on the right cues, when what has been established is that the cues we labelled were followed by good outcomes. Those are different claims, and the second one does not survive the coding process that produced it.

I would rather see one honest figure than any amount of trigger taxonomy. The proportion of qualifying cues on which the press was launched, and the proportion of launched presses that recovered the ball within a stated number of seconds. Two numbers with defined denominators.

None of this means pressing analysis is worthless. Coaches who work with video have real insight and the eye test is not nothing.

But once you write it down in a spreadsheet it acquires an authority it has not earned, because the spreadsheet does not record that the rows were chosen after the endings were known.