Broadly you’ve got it. Basically misfit risk allows the manager to pursue their style. There’s an optimal amount of misfit risk, however it isn’t always optimized at zero. Therefore, by pursuing their style, they can generate higher alpha.
I do not believe so. Take an example. Suppose my benchmark is US small cap. I decide to implement this with a small cap value manager and a small cap growth manager (maybe not in the same proportions as the overall benchmark). When evaluating the small cap value manager, they should have a small cap value benchmark that they are measured against. Their tracking error will be relative to this value benchmark. In evaluating the overall portfolio performance though, I have a tracking error relative to the overall benchmark - the small cap index (no value or growth tilt/bias). The difference between these - basically between the small cap benchmark and the small cap value benchmark is the misfit risk. In evaluating the manager, I can’t/shouldn’t compare them to an index that is not consistent wit their style.
Think of breaking down the return between what the manager does relative to their benchmark (real active retrun/risk) and what the overall portfolio manager chooses in implementing the overall strategy (the “misfit” return/risk). The manager should not be credited or deducted for the choices made at the portfolio level, thus the term misfit. (which can be confusing, because it’s a misfit for the manager but in fact an intentional act at the portfolio level typically as part of the TAA)