the linear format of player role formulas makes them insufficient to capture the complexity of player abilities. you can use them to put together a decent guide though, to aid with the comparison process. for example, great per is not worth much on a guard with poor spd and bh, but it is incredibly valuable when paired with good spd and bh - the variable nature of the importance of one rating, with respect to other ratings, means you cannot accurately capture the value of a player with a simple linear formula with coefficients. however, you can safely assume per is at least pretty important in your SG, so you can pick a value that isn't that far off, for SGs who are very per dependent, and for SGs who are less so. in short, player roles can be a useful tool, but never expect to be able to replace the eye test.
player ratings are useful pieces of information - but coaches should really strive to see players in terms of their abilities, not in terms of their ratings. how good of a shooter is the guard, how good of a scorer overall, how good of a defender, passer, etc... that is what matters, its the aggregates that the game runs off of, that we need to try to understand. the game absolutely takes a player and converts them to a number for defensive ability - if we aren't thinking about it that way, we aren't going to be in sync with the underlying mechanics, and our projection and models will be flawed. abilities are the true building blocks of a player and team - and a key job of the coach is to be able to map player ratings to abilities, and to map a set of player abilities to team abilities. understanding the essence of what makes a player good, and what makes a team good, is far more important to success than most people give it credit for.