The question you ask, I think is fair and good, and I wish you to find the right answer. Probably the question is in the mind of many young researchers, although the answer should be individual, I think it is relevant for MO. But actually I am not in good position to give advises on your main question,
so let me comment on the following " ...bet on Bigger results, which may not succeed..."
I do not think it is good to "bet on Big results", I mean you may try to invest your time to studying some big conjecture or modern popular technique, but when one is young it is difficult (imho) to estimate properly ratio (efforts spent/result obtained).
The way which I choose for myself (actually much later when I got PhD) is the following: I start working on the subject when:
a) I am 100% sure that I will get "some" (may be "bad") result, which can be obtained in reasonable amount of time - say 2-3 months
b) I hope for some "big result" if I would be lucky/clever/whatever. I mean I see some "big problem", such that 100% guaranteed result is some "partial case" of it.
To the best of knowledge similar way is used by many people.