Overall of D. Paraschiv in every position. In his own position it is EA’s real rating; the rest are estimated from the individual attributes with a per-position weight vector calibrated over 13785 players, anchored to that real value.
Method: the rating in the player’s own position is EA’s real value (exact). For every other position we apply, on top of that anchor, each position’s weights — derived by linear regression (OLS) over our ~18k players, so they self-calibrate every edition.