Tinstaafl sums it up very well and is an abridged version of John Ecklabars comments in his book.
The angle of bank for performance as opposed to Vmca issues is often in the order of 1.2 - 2.4 degrees for a light twin, more for more powerful machines, (B58) 2-3 degrees.
Interestingly the theory outlines that for every degree beyond optimum 30'/min ROC is sacrificed. The importance of not over cooking the angle of bank whilst targeting the correct angle is clear. I doubt many of us could guess accurately what is 2 degrees when under the pressure of an early engine failure, however knowing how important it can be might lead to a slightly better outcome.
I should add that for Vmca issues at least 5 degrees is the number to strive for and that for every degree less than 5 degrees the Vmca will rise by 3 kts this is based on data from a Cessna Conquest apparently. So anyone reading this please understand that there are two issues here, controllability and performance and different numbers for optimum on each.