"observed" meaning that we making tests with the same (as possible) conditions and from the records, we have some evidence -conclusions.
From lv3 I 'm keeping records about training and fast trainers. Here is a picture from a file of one season- it's the training and the board with gaining and rates after (almost) every game I had that season
but here we have an equation with known values (stars and quality, age, morale & condition, Value & Wage, types of training, previous team) so I figured out the unknown x, the gaining ability.
Still the are sometimes that it gives different values. eg a 5* nordgen, in same conditions can have a +90 or a +120 which is a big difference. After "observed" the conditions, I figured that it was just random, a roll dice that can affect the gaining, but most of the times the value and the average have a very steady pattern.
So my friend Amane, I wonder in which way you 're making experiments ? I don't underestimate your knowledge about the game but how you figure out about captain behavior ? I can only think of playing 10 (friendly) games, with the same team, with same conditions and using 5 times one player and 5 times another.