Well, what do you mean "way" behind. A hypothetical big with all 8 guard skills is generally considered well rounded, but 8 is still *WAY* behind, say, a 17 or 18 inside skill. (E.g., the standard deviation among skills on a 8/8/8/8/8/8 - 17/17/17/10 guy is HUGE, so even though we might consider that guy "well rounded", a number of possible algorithms would see that guy and think mono-skilled monster). But this isn't the right thread for it, and perhaps based on Marin's comment we were seeing something unusual, not typical of the changes.
But unless I'm wrong, no one knows what the "worst-case scenario" is for cross-training, right? We know the average effect is 10%, but we don't know whether the range across players is 8-12%, or 5-20%, or 0-200%.