Schedule has to be built around the pilots returning home as the flying crew at that particular base. It's inefficient. The optimal schedule would be pilots stick with the airplane forever. In the Part 135 world, this is how it works; if they crew swap, the company buys a ticket or a rental car to reposition. The fact that the schedules have to be built around certain people returning to a certain place means it makes the overall operation less efficient. Deadheading in Part 135 isn't really an issue because the charter rates include all these costs. Deadheading on company means less potential revenue per seat used. The dollars spent per night per pilot for hotel rooms and per diem are far less than the reserve pilots you'd have to keep around at that domicile. Now that pilots can 'live' in that base, they can drop trips, call in sick, personal day, etc. out of that same place. An overnight really only risks crew timing out, or severe sickness, in which case one reserve from one base is deadheaded out there. Now consider that the cost to the company per employee is like 50% more than they pay him, with their share of FICA taxes, health insurance, training, etc. you can see that hotels are, for an accountant, a far easier cost to quantify, and less than a domicile. There'd be no domiciles in a perfect world from management/accounting standpoint.

DL has more nonstop routes on the A330 type from CDG than it does on the A350 out of LAX. But you could obviously see how it would cost a lot more to make CDG a base. Same reason Hawaiian doesn't have a mainland base. Extreme examples, sure, but it proves the point.