What, then, of attributes? How may they be used to assess a player?
I think attributes can best be thought of as a record, in numerical form, of a player’s current capability.
They’re particularly useful for comparative purposes. A player with a pace attribute of 10 is currently capable of running faster than one with a corresponding figure of 9.
They are rendered more useful by the fact that they tend to be stable over the short term, unlike form – which can alter dramatically.
How long is the ‘short term’? That depends, in part, on the age and experience of the player. Play a 16 year-old in half a dozen successive first-team matches and his attributes are likely to start increasing noticeably. The same is unlikely to happen with a 24 year-old. (The attributes of older players who are in decline, however, can fall away rapidly.)
Attributes are dangerous things. Because they are typically presented in numerical form (and, to boot, as integers), they look comfortingly unambiguous and reliable.
In fact, though, they need to be handled with care. They require interpretation.
For one thing, they indicate only the limits of capability. That makes them weak as predictors. The player with a pace of attribute of 10 could run faster than the one with 9 – but might not. It depends – for example, how tired they are, what their morale is like, how confident they are feeling, and so on.
Which leads to a second point: that attributes make most sense when seen in combination. Will a player get to the ball? That will partly be a function of pace. But it will be a function of many other attributes to – for example, whether they are in the right position, how well they have anticipated the situation, and whether they have sufficient bravery and aggression to throw themselves into a challenge.
Reading attributes, therefore, is not simply a matter of taking individual readings, as with a tape measure: it is also a matter of recognising patterns.
Overall, attribute figures really have no value except as predictors, yet their reliability is dependent on (a) other attributes, (b) non-attributional phenomena (such as personality), and (c) change over time.
I will need constantly to remind myself that if I treat an attribute as the oh-so-simple datum that at first sight it appears to me, it will trip me up.