When you measure decides what trend you find

A testing schedule tied to convenient days samples the season unevenly and produces seasonal patterns out of scheduling.

Tuesdays, before training, because that is when the room is free. Innocent enough as an arrangement, and it can generate a season-long trend that has nothing to do with the athletes.

Tuesday is not a random day. In a one-game week it is two days after a match and the squad arrives sore. In a two-game week it may be a day before travel, with the session shortened and everyone fresher. Across a season the mix of what a Tuesday means shifts as the fixture list bunches and thins, so the conditions under which you measure change systematically across the months while the protocol stays nominally identical. Any drift in the data will be read as adaptation. Some of it is the calendar.

The same trap operates within a day. Tendon and muscle properties are not the same at nine in the morning as at four in the afternoon, and the testing slot in most buildings moves around depending on what the coaching staff needed. Half your preseason readings taken at eight, half your midseason readings taken after lunch, and there is your seasonal change, generated entirely by the diary.

What makes this hard to catch is that it produces smooth, plausible curves. Noise looks like noise and people discount it. A gradual drift looks like biology and people explain it, usually with whatever training emphasis happened to be running at the time. The explanation arrives faster than the check.

The check is to record the context, every time, in the same file as the number. Hours since last session. Time of day. Days to next match. Then plot the metric against those columns before plotting it against the date. If the reading depends more on hours-since-training than on the month, you have learned that your monitoring is measuring the schedule.

Missing sessions make it worse in a specific way. Players are absent from testing for reasons connected to their condition: injured, resting, away. So the composition of the group changes week to week, and the squad average can rise simply because the three sorest players did not attend. A group mean computed from whoever showed up is not a time series of the same thing.

Before anyone interprets a trend, I want to know who was in the room each week. That list explains more curves than the training programme does.