There could be a penalty in performance depending on position and playcall. It affects some positions more and some less. RB is a position where scheme matters less and we all know the reason. If you have a fast RB he will be effective. The penalty in performance would come when you got something like a cover 2 corner and try to play man defense. This happened to me. I was in a online CFM and got stuck with the Vikings roughly two years ago. I use the Ravens. Believe me relying on Winfield in man coverage cost me. In general it affects defense more than offense.
No, it is a problem with how speed, acceleration, and agility are nearly the only things that matter for RBs. And forgive me but if you cannot even recall what scheme you used it does make me question exactly how much you payed attention. Aside from that, I already granted that for RBs it does not matter quite as much unless you call specific plays. What running plays did you call? But if you think a player should run slower, be less agile, etc simply because he is playing out of scheme then we have reached a fundamental disagreement.
No it is not simple math, because there are different overalls. The scheme influenced overall is not the real overall, i.e. it is not the overall driving the engine. In this case the scheme overall is shown only for your benefit. But the under the hood mechanics that determine player salary is driven by the baseline overall that is displayed on the front end. Madden just released all the ratings, now if you start a CFM and change scheme that will result in overall changes, but the game still calculates salary based on the overall that was recently released.
I actually do disagree with this statement. The problem is not with scheme but with the player ratings and their impact on gameplay such as the aforementioned speed of RBs. And while schemes were introduced partly to improve AI roster management, there are still problems in the roster management code and I think that is separate from schemes. Tied to this is the small rating scale for important attributes. Take route running for example, the range for WRs s quite small. The majority of WRs have RTR in the mid 80s. So there is not much variation to make route running WRs a viable option.