I hope it's acceptable to discuss just one philosopher in my answer because he did a lot of thinking about these issues. W.V. Quine sets out six virtues of plausible hypotheses: conservation, modesty, simplicity, generality, refutability and precision.
These are virtues (ideal hypotheses), very much to Quine's taste. Which is rather funny since Quine likes to demand empirical evidence for everything, and I don't know if all these virtues can be supported by empirical evidence. They sound suspiciously rational to me, that is, they sound like they were generated wholly by mind. And even worse for Quine, they sound like just matters of taste!
In the real world, I think scientists get their hypotheses from many areas, dreams, hunches, etc., and of course there are fortuitous (read, lucky) accidents in the lab that result in great new discoveries and theories.
I guess a scientist could whip-up a beautiful, virtuous hypothesis after the fact to satisfy Quine's six virtues.
Nevertheless, I think Quine still has something to offer to the scientist when he speaks about theories and other things. Theories he says are underdetermined by the evidence because when we reduce them to language we extend them, so that not all of the theory is supported by empirical evidence.
Now even though our sound scientific theories are underdetermined, they still can be a fact of the matter. He specifically mentions physics in this connection.
Why are they a fact of the matter? " Because good theories are the simplest theories that account for all the evidence. These entities and their properties and interrelations (which make up theories) are all there is to the world, and all there is to be right or wrong about." Dagfin Follesdal on Quine.
So I would say that for Quine the good theory would be based on empirical evidence, it would be as simple as possible but still take into account all of the evidence.
To Devanil, then I would answer that the hypothesis in the "real world" might be even a guess, a hunch, the scientist has an idea and this sends him to the lab to test it! Regarding theories, I follow Quine in thinking we need empirical evidence for them. However, when we talk or write about theories we are likely to extend their claims beyond what the evidence strictly allows.
Finally, there are facts of the matter in science, but not in other things we study (Quine would say). Of course this does not mean that scientific theories cannot be revised. Certainly they can. All Quine is suggesting in that science (and specifically) physics can offer a fact of the matter, at least for the time being until such a theory may be revised.
Ref: Gibson, Roger F. Jr, (1988), Enlightened Empiricism: An Examination of WV Quine's Theory of Knowledge, U South Florida, Tampa. Great book that pulls Quine's scraps together and makes sense of them.