The combat system is far to complex to allow one person have a coherent perspective on all class balancing. That's why BW has relied so much on its in game stat collection system. It was supposed to fix the need to have a lot of human labor hours put into class balancing. Needless to say, the statistical underpinnings of the system were flawed.
The problem for BW is that the conventional methods of involving the player base are also badly flawed. Look at these forums. During PTS testing the class gripe thought to be the second most valid gripe was by Assassin dps who believed they were nerfed (coming in second to the Merc dps griping as the most valid it was supposed). But now look at what is actually transpiring. Assassin dps is hands down the best single target pvp dps class. Better than Vanguard/PT. Better than Operative dps. Only Sniper dps can compete and then only vs. a melee heavy opposing team. Bottom line - the devs can NOT rely on forum griping to provide an accurate look at class balance.
Fortunately there is a solution. One that uses BW's automated data collection. And player involvement. Yes, it's true Virginia.
THE SOLUTION - measure player class useage when they win. This automatically corrects for the problems with the current stat collection system which overweights bad players, particularly in pvp. So for pvp, BW needs to collect data on the subclasses used by each team that wins a ranked pvp match. If a player uses a tank class in one half of a Voidstar and a dps class in the other half, half credit should be applied to each subclass. Then simply count which subclasses are used most often. Those are the subclasses that are OP. Those least used are underpowered. Ignore the normal wz matches. The subset of data provided by winning ranked wz team composition is the best proxy for which classes the best players believe are most effective. It avoids the problems inherent with damage or K/D ratio analysis that can not account for classes used as defensive guards or offensive scouts. The only weakness in using data this way is that it doesn't give a good view of tank v. healer v. dps dynamics. But it is the BEST way to look at balance between the three tank classes. And between the 3 healing classes. And between the dps classes.
For pve the same methodology can be used w.r.t teams that finish the final boss in the hardest operations run available. These are the best pve teams in the game. Measure the success. Stop measuring the average players and start doing what should have been done from the start - manage class balance based on best practices for each class. Which is what your are approximating by seeing which subclasses are winning.
I'm ArchangelLBC and I approve this message.
Although the shadow pvp dps being so high is a direct result of their gripes being addressed (and perhaps overaddressed).