From a shoryuken.com discussion thread.
cjaycr: “What exactly was [Capcom’s] approach to balancing the game?
Their approach was a hybrid general system adjustment and general character adjustment combined with a very SNK-flavored character rework element.
At least, that’s how it looks to me.
Unfortunately, the team was too passive on some aspects and too extreme in others. This sounds like “par for the course” since we’re all human, but most of the extreme elements of the game stem from the game’s very serious mechanic flaws.
General System Approach:
The general system approach seemed very positive for the most part. Meter gain reduction? I think we can all agree that was a good way to go. XF1 nerf? We were complaining nonstop about XF1 Wolverine–he was practically the tournament mascot for the game. Air dash adjustment? Stops that absolutely brainless rush that all of us (ALL of us) were guilty of at some point or another. TAC rework? Okay, this is where people get mad, but I think they did the best they could with this without straight ripping TvC’s TAC (which they probably should have done in the end anyway). They took what was essentially a bad RPS style damage/meter mechanic and turned it into a hyper-strategic RPS meter-centered mechanic. While most players are quick to dismiss “hyper-strategic,” what I mean by that is the notion of certain characters wanting to drain meter and certain characters being hungry for it, making the guessing more meaningful and some mid and low-level play more obvious. Unfortunately, yomi doesn’t stack in this game very well, so since there is a third WTF choice in there for players to pump-fake with, the choices become more weighted on both sides. The TAC fix isn’t really a fix, though, because….why? Well, TAC is still inherently poorly designed. A flagship mechanic isn’t going to get removed from the game, though (see: X-Factor).
General Character Approach:

The general character approach was an example of mostly positive, but extremely passive adjustments. No real broad adjustments to minimum damage scaling anywhere. Most characters got some new toys, some necessary (Spencer’s Bionic Bomber, Hulk’s command normal, etc.), some unnecessary (She-Hulk’s command normal, new Hsien-Ko items, etc.) and some just for fun or to give the characters something new (C. Viper’s optic blast, Morrigan’s meter fireball, etc.). Some characters got faster air-dashes or a minimum scaling change or increased pushback or float on some things…very patch-like changes to what was essentially a new game coming out. There aren’t too many really -exceptionally- positive buffs passed around to make an entire point around, but most people focus on the negative anyway, so let’s check it out. Zero’s cancelable level 3 buster, for one. What about really negative nerfs? You only have to look at She-Hulk. These two examples show what happens when you make adjustments based on a broken meta. Zero probably wasn’t on enough peoples’ radars (for how prominent the character was probably designed to be) and She-Hulk was just seeing too much success, so Zero got something to make him more appealing (and much more powerful) and She-Hulk got the bump. Some of their changes are just confusing like the Haggar buffs despite his assist nerf (which I think was a great example of a full character buff without reworking the character) compared to the Tron nerfs COMBINED with her assist nerf (which I suppose is a great example of a full character nerf in almost every dimension without taking them out of the game…or reworking them).
Character Rework Approach:
The real interesting bits are the character reworks. Reworks, for those who may be confusing the term with nerfs and buffs, are character overhauls that are meant to change the way a character is played or to at least force a different style of play by changing broad fundamental aspects of the character. Again, we have more examples of passively positive reworks combined with some confusingly questionable reworks. Wesker is my favorite example of a character rework in this game. Aside from Thor and Magneto, Wesker was one of the heaviest reset-based characters in the game and, if you look at it from a certain angle, the only one (nobody played Thor and Magneto got more bang for his buck by finishing vanilla damage combos anyway). The problem was that Wesker got too much out of being a simple reset character; the health, the hitboxes, the one frame command grab, the teleports, all that combined with the fact that combos brought in way too much meter and resets meant never having to spend any of it…everything was too big of a package. I would imagine the desire here was to change (I should probably say dilute) Wesker not by the numbers (damage, meter, scaling, etc.), but by playstyle by making it more appealing to finish combos and spend meter in order to take his glasses off. This, to some, might not seem very significant, but what do we see and hear over the mics at majors everytime Wesker comes in on a full team in the anchor spot? “Okay, all he needs is a touch and he can set up.” Wesker gets a combo and quickly (many times not even optimally), goes straight into…Phantom Dance. “Okay, now he’s got his glasses off.” We’ve heard it almost every other Sunday evening on streams or at majors for the past three years. While it’s up to players to decide, I think the Wesker rework was headed in the right direction; I think they were too passive on some of the things they reduced and too forthcoming on some of the things they added, but the real intention to change Wesker for the better is evident.

I’ll wrap this up with the most unfortunate character rework in the game: Iron Man. Poor Tony. Does anybody remember the tagline for Mr. Stark during the days of vanilla Marvel? “He’s just a poor man’s Magneto; everything he does, Magneto does better and easier.” As such, there was no real reason to play as Iron Man unless you wanted the Unibeam assist. I can only imagine that the team picked up on this and said, “How can we make Iron Man noticeably different from Magneto and Storm? How can we give him his own niche to fit into?” To Tony’s chagrin, in comes Ultimate Iron Man, who was reworked with the intention to change from a heavy tridash rushdown character that had projectile and reversal options to become a defensively-oriented projectile character with emphasis on half-screen hitconfirms. Tony’s new gameplan doesn’t revolve around his tridash anymore; indeed, it can’t! He’s forced to make more use of Repulsor Blast and now gets much more mileage out of Unibeam. Tony is actually supposed to hang back now and essentially whiff punish with his M and H moves into decent, but unusually fragile combos with Proton Cannon as a backup reversal option. He gets most of his damage in the corner with multiple uses of Repulsor and he gets to the corner with his new dash-cancelable aerials. Was the character reworked successfully? Yes. Was the rework a success? Probably not. Tony changed for the better in so many ways, but the fundamental change to his core mechanics means that his old style of play or the desired style of play isn’t the optimal was to play the character by design. Can you still rush with Iron Man? I mean, you can.
So that’s it. I think the evidence is in the game of some really positive changes, but when you look at logic like “We kept Hsien-Ko down because we didn’t want gold armor to be too busted,” and consider that gold armor is still risky ANYWAY, you can see that the follow-through wasn’t there to coincide with the good intentions. Lots of people get mad at the way the game turned out and it’s understandable, but the changes aren’t as inane as they seem. They’re just:
1] mostly poorly done
2] adjusted around a busted meta and
3] implemented in a game with fundamental design flaws and questionable mechanics.