If I'm understanding the terminology correctly, in older versions of XanderCat I used cropping in the gun (using basic MEA calculation as a base but only shooting where my predictor claimed the opponent could reach). However, in the latest versions I use scaling in the gun (maximum reach based on the predictor mapped to the end bins). I do not use scaling in my wave surfing drive, however. Others results may vary, but in my robot, scaling improved performance in my gun but reduced performance in my drive. For what reasons I haven't fully analyzed.
And of course, then there is the question of how the bins are scaled across the MEA range. I did a little drawing on some approaches on a precise MEA discussion. For my gun, I currently use evenly distributed bins (image 3 of my awesome MSPaint diagram on the Precase MEA discussion page), which has the interesting aspect that the 0 factor location for me is not fixed. I have not fully researched the effectiveness of differing approaches on this.