Well, I tend to use more than strictly power to define how I build something. In dealing with 40 year old castings, I find it usually necessary to either replace seats, or use oversized valves in order to get the actual valve seat angle back up closer to the chamber. I'll seperate here into intake and exh sides..maybe that would help.
On exh sides- I install hardened seats, so seat angle placement is basically irrelevant. Very generally, a larger valve size will give you more low lift flow potential, because the bigger diameter means more flow area for any given lift than a smaller valve. Also, as CP said, pushing the seat out (towards the exh port opening) on the exh side means a larger radius on the short side turn, which is a big part of an exh port's ability to move decent numbers. It also means a cooler running exh valve, because it's transferring more heat to the head when it's closed.
On intake side - I will many times install the larger valves just because of the seat condition. I'll get more power from a seat placed properly than I'll lose from "low velocity due to a bigger valve". I quote that because it's common to hear, and can be misleading. A very fast port can benefit from a larger valve, because it helps the air slow down and make the turn around the valve. A large port (volume wise) that typically has lower low speed (rpm wise) performance can be hindered at those low speeds even further by too large a valve. Where an almost stock factory port may benefit slightly from delaying the point (lift wise) at which the port goes turbulent from too much velocity. In most ports, if the pushrod pinch is not modified, the valve size is not as important as most think. Because the "minimum cross section" will always be at that point in a stock head intake port. Lastly, on the intake, if its a 383 or smaller and the bore is less than 4.28 (383+.030) you will lose a lot more from the bore shrouding the intake valve. Many head shops will flow test on 4.5" bores, because it yeilds the best numbers. (hence the warnings about not racing flowbenches...) Quite a bit higher than what the head will typically flow on a 4.25 or 4.35 bore. It's the bore shrouding that lessens the value. And a smaller stock size valve (2.08) will flow much better on that type of engine. A "rough guide" for intake valve size is do not ever exceed 52% of the bore size on a race engine. A street engine (lower operating rpms) I like to stay at a max of 50%. So a 2.14 would be my choice over a 2.18 on a 440. Where 2.08 would be first choice (if seats make it possible) on a 383. If you're buying valves anyway, size doesnt matter. But if you have to buy and install 8 more seats, that's hundreds more.
Every builder has his own feelings on this. So I'm sure there are all kinds of opinions. But this line of thinking works for me...