Since the question was "what size", not "do I need" ...
I do not subscribe to the idea that it is an extra point of failure. In a chain, every link is a point of failure. I am least concerned about the swivel, in that chain of vulnerabilities....
I am wondering if they just anchor in tamer conditions ....
or if it really doesn't matter for a day or two on the hook.
Points worthy of discussion.
Extra point of failure. Regarding strength, if you match WLL there should be no issue. But there are two additional factors to throw in the mix.
- Inspection. You can look at a chain or shackle. The failure points of a swivel are most often hidden from view. The crack on mine I only saw when the shackle was disassembled, since it was obscured by parts of the shackle itself.
- Leverage. Many swivels are installed dirrectly to the anchor shank. Indeed, they are designed and advertised that way, even though it adds a terrible side load to the swivel and a side load to the anchor. Mine was designed this way. I think most of us would agree this is a serious mistake, but some of the readers of this thread may not have thought of this issue. There should be a few links of chain or a pair of shackles between the shank and swivel to eliminate this leverage.
Tamer Conditions. I'm pretty sure this has nothing to do with the debate. I've never heard it argued that a swivel improved holding in a storm.
A day or two. A valid point. I've never anchored for more than 7 days (I had a cruising cat for 15 years). I moved on, and I think this is true of the vast, vast majority. If you live on the hook, will stay in one place for weeks, and the wind or tide will 360 the bot daily, yes, a swivel is needed just as it is on a mooring. I think most would agree with this. The anchor is a mooring.
But in 7 days, how many times will the boat turn a 360? Probably not twice every day, since some turns will be in the reverse direction. Some times it won't spin at all, because the wind is steady and the tidal current is negligible (not all tidal harbors have a measurable current). So at worst, the boat spins 10 times in 7 days. That's probably less than 1 turn in 10 feet and will not be an issue. It will spin out. But for most of us, a typical few-day anchor might mean 2-3 turns at most. So yes, it depends.
Life expectancy of stainless. Galvanized in re4gular use lasts 4-8 years, depending on use. Stainless looks good forever, but in saltwater its fatigue/corrosion/cracking life is about the same as the life expectancy of galvanized. If stainless components are replaced at the same frequency as their galvanized cousins, all would be well, but typically that shiny SS swivel might get reused. It should not. My cracked swivel was probably 15 years old, probably on its second chain.
----
What actually fails? Failure of chain, unless in truly horrendous rusted and worn condition, is unheard of. Shackle failures nearly always involve unscrewing; tighten them, mouse them, and perhaps add Loctite. Quite avoidable. Swivel failures are not unheard of, but most are either old (more than one chain), cheap, or attached to the anchor shank. A rope rode can cut, but you don't often hear of it. The rope can chafe on the roller side plates; more likely, but completely avoidable. Mooring pendants chafe through with every major storm. Or the anchor drags. Obviously, 1000 times more likely, and the reason is seldom any type of gear failure.
As I said, I removed mine and never missed it. In fact, because the windlass and roller keep the chain aligned, I had fewer backwards anchor incidents after I tool the swivel off. With the swivel there was a 50/50 chance the anchor was reversed. Without the swivel the anchor would come up right-way 98% of the time, because I drifted backwards and because the chain wanted to come up twist free, aligned with the windlass. That is my expereince.
If I stayed in one place long enough that my anchor was a default mooring, I bet I would have a swivel.