The transition clock starts a second and a half after the transition began

Timing counterattacks from the moment possession is officially recorded misses the seconds where the outcome was already decided.

Five seconds from turnover to shot. From which turnover, though?

Event data marks possession change at the point the new team has clear control of the ball. That is a sensible convention for bookkeeping and a poor one for measuring transitions, because the decisive part of a counterattack often begins before that timestamp. A loose ball is contested, a striker is already running, a midfielder has already turned. By the time the log records a change of possession, the shape of the attack has been set and the defence has already lost or won the argument.

So the standard transition duration is measured from a start point that arrives late, and it arrives late by a variable amount, which is worse. In a clean interception, the recorded start is close to the real one. In a scrappy sequence with two ricochets, it can be a second and a half adrift, and a second and a half in transition is an enormous quantity of pitch.

That variability means transition timings are not comparable between the situations people most want to compare. Counterattacks from turnovers in midfield scrambles will systematically appear faster than they were, because part of the attack occurred before the clock started.

There is an equivalent problem at the other end. The transition is often deemed to end at a shot, which means sequences that create a chance without a shot, or that end in a corner, drop out of the sample entirely. Selecting on shots again.

Tracking data can fix the start point. You can define the transition as beginning when the collective velocity of the two teams reverses, which is measurable, continuous and does not depend on anybody deciding who has control of the ball. That definition would move a lot of published transition numbers.

Nobody is rushing to redo it, partly because the current figures support a well established narrative about how quickly the best counterattacking teams operate, and revised figures would disturb comparisons that people have already built arguments on.

I have some sympathy. Consistency has value even when the definition is flawed, provided the flaw is disclosed.

It is never disclosed. The number appears as a clean interval, five seconds, and nothing on the graphic tells you that the stopwatch was started by a convention rather than by the event it claims to time.