Everybody who measures a watch for the first time wants to know the same thing, which is whether the number they got is any good. It is a fair question with an unsatisfying answer, and the unsatisfying answer is more useful than a tidy one.
The bands, and what they are for
Two reference bands come up repeatedly, both quoted in seconds per day.
−4/+6 allows a watch to lose up to four seconds a day or gain up to six. 0/+5 is tighter and one-sided, permitting a gain of up to five seconds a day and no loss at all.
These are the strict end of the scale. They exist as testing standards, applied under controlled conditions to movements built and adjusted to meet them, and a watch held to one of them is a watch that somebody took trouble over.
What matters for your purposes is that they are reference lines rather than pass marks for watches in general. A great many perfectly good mechanical watches are specified far wider than either, by tens of seconds a day, and are performing exactly as intended at figures that would fail both bands comfortably.
So what should you expect
If you have no specification to hand, the useful frame is not a threshold but a range of outcomes.
A mechanical watch keeping time inside a handful of seconds a day is doing well by any standard. One in the region of ten to twenty seconds a day is ordinary, and describes an enormous number of watches that their owners are perfectly happy with. Beyond that, the figure starts to be something you notice in daily life rather than something you have to measure to find.
None of those are verdicts. A watch running at twenty seconds a day may be exactly as its maker specified. A watch running at three may be about to need attention. The number on its own does not distinguish those cases, which is the whole reason the next section exists.
Consistency beats zero
Most accuracy discussions skip this.
A watch that reads +6 every single time you measure it is a better-behaved watch than one that averages +1 while swinging between −9 and +11. The first is predictable, which means it can be regulated closer to zero by somebody who adjusts watches. The second is not, and regulating it would only move the middle of a wide range.
So when you look at your own readings, look at the spread before you look at the average. A tight cluster is the good sign. A number close to zero that arrived by accident, from a watch that gains hard on some days and loses hard on others, is worth less than it appears.
This is also why one reading answers very little. Seconds per day covers what a single figure does and does not contain.
What the conditions were
A rate is always a rate under conditions, and the conditions can move the answer by more than the difference between a good watch and a poor one.
The same watch, measured across a week on the wrist, across a weekend in a drawer, and across a night resting on its side, can honestly produce three quite different figures. Gravity moves the rate depending on how the watch is lying, and a watch running down toward the end of its reserve behaves differently from one near a full mainspring.
None of that makes the measurements wrong. It makes them answers to different questions, and it is why comparing your figure with somebody else’s figure for the same model is mostly a way to confuse yourself.
What the number cannot settle
A band describes daily rate and nothing else. It has no view on how a watch handles water, a magnetic field or a knock, and it is not a report on the movement’s condition.
Sitting comfortably inside one is no guarantee that a watch is not overdue for attention, and sitting well outside one is no evidence that it is. The second watch may be in fine health, adjusted to a different taste, or measured across a week that was not representative.
The rate is one property, measured one way, and genuinely useful for what it covers.
The question worth asking instead
Rather than asking whether your number is good, ask what it does over time.
A watch that has run at +5 for a year and is still running at +5 has told you something reassuring, whatever you think of the figure. A watch that ran at +3 for a year and now reads +25 has told you something worth mentioning to a watchmaker, without telling you what it is.
That is the version of this question a rate can actually answer.