One of the most interesting things about covering one subject for years and years and years aside from wondering increasingly if writing about wristwatches all day really counts as a suitable job for a responsible adult is that you start to notice how many gaps there still are in your own knowledge and how little you've actually fact-checked things you've repeated as conventional wisdom for all those years.
Not long ago, I got to talking with someone who in the line of duty hears, much more often than I do, from customers who sometimes feel their expectations for accuracy from their watches are not being met; this person asked me what I thought a reasonable expectation was. There is not, I think, a single answer to this question; it really does depend on the watch and it also depends on the individual owner. In many cases, whether your expectations are grounded in reality can simply be checked against the manufacturer's stated spec for the watch in question.
In both cases you may get better performance; it's been my experience anecdotally that Grand Seiko routinely underreports the accuracy of their watches but I certainly don't have a statistically significant sample.
A friend who bought a Grand Seiko GMT, non-high beat, a month ago says he hasn't noticed it being even a minute off; a minute per month maximum gain or loss was once advertised as a major achievement in the s by Girard-Perregaux, for their Chronometer HF high frequency watches.
Beyond manufacturers who publish their internal accuracy specs, and those who have their watches chronometer certified, things can get a bit harder to pin down. However, you can get an idea from the specs published by ETA or Sellita for their different grades of movements.
