I enjoy the Confidence system as-is for a number of reasons as a game mechanic and I really don't think that it is unbalanced when combined with the ability to progress the Consistency rating and the ability to increase a player's confidence with Game Prep. However, I'm not going to focus on what I like about it in this post, rather I'm going to talk about how I think the whole thing could be made better.
I do think that the Confidence system - and the entire game, really - ought to more directly consider context in which stats were accumulated. For example, if a player makes one play in the entire game but it's the game-winning play (be it a TD, INT, sack, 30-yard run which sets up a game-winning field goal, whatever), that player ought to get at least a small boost and the "I should have contributed more!" penalty should be cancelled. The player just won the game, he should be pumped! He came through for his team when he was needed! Conversely, if a player runs for 200 yards, that shouldn't mean much if he lost a fumble which ended up losing the game for his team. Or if you throw for 500 yards but throw a game-losing INT on the final possession a la Tony Romo vs Denver in 2013.
In general everything in Madden CFM is very stats-driven, which while fine in previous years (particularly in a competitive multiplayer setting). I've been very vocally supportive of the current stats-driven gameplay loops of CFM. However, primary stats alone to drive the mode may no longer be sufficient going forward. Primary stats such as yards and touchdowns don't capture the narrative ebb and flow of a football game. A player running for 150 yards and two touchdowns means nothing without context; did those two touchdowns directly help his team win a la MVP season Shaun Alexander, or is the running back just grinding away for yards on a bad team with no other options a la Emmitt Smith in the early 2000s?
A potential solution - tracking win probability over the course of the game could help solve this issue. With that,
Win Probability Added could be aggregated per-player, and the player's WPA could be used to mutate any modifications to confidence (and also XP derived from statistical achievements). Click the link for a more thorough explanation, but the idea behind WPA is to figure out how much a player contributed to a hypothetical victory or loss for his team.
Using WPA values calculated by Advanced Football Analytics from the recent
Cowboys - Rams game in real life as an example, we can see that even though DeMarco Murray ran for 100 yards, he still had a fumble early which put his team in the hole, and that's accounted for as his WPA is slightly below zero. Rams' TE Jared Cook also ends up with a negative WPA on account of his dropped touchdown pass, even though he statistically torched the Cowboys; he missed an opportunity to really help his team seal the game. Tony Romo ends up with a huge WPA - even in spite of his pick-six to Janoris Jenkins - because the INT didn't contribute much to the Cowboys' likelihood of losing the game (the Cowboys win probability moved from 0.09 to 0.07 when they went from being down by 14 in the 2nd to being down by 21 in the 2nd), and every play Romo made going forward from that point contributed positively to his team's chances of winning, all the way through the completion of the comeback until the Cowboys' win probability is 1.00 (when they won the game and the clock expired).
I realize that's complicated and probably too much thought for me to be putting into something at 2 in the morning, but I digress. I've been thinking about WPA a lot lately and how it could be applied to Madden, and calculating changes in player confidence is just one spot where I think it could be a useful metric to reference.