Both, until I feel like I've hit upon
the strategy, then I just keep examining different ways to perfect the one general strategy. In games, there are usually 2 central components of decision making
1. Risk and reward analysis. This is so so so so important. Whenever you make a risk, your rewards have to be at least equal to that risk. Like in poker, it doesn't make sense for you to go all in with 5000 chips when there's only 2000 in the pot. What if you get it right? What if it goes wrong? You need to think about all that. Furthermore, if there's a low risk move with a high reward, or one that pretty much guarantees progress in some shape or form, you pretty much always take that route.
2. Your read. Based on your game's rules and game mechanics, there are only so many reasonable moves your opponent can make. Remember that they should be analysing their risks and rewards too. Your own moves should hence be able to cover your opponents. This doesn't necessarily mean that you need to react to them, just don't put yourself into a bad position given their options. Of course, if you can take advantage of your reads, in a way that is decent in terms of risk/reward, you should go do that. Though prediction is important, I also want to stress that it is still ultimately
a guess. There is no such thing as knowing what your opponent will do 100%. Anyone who claims that literally is either fooled or lying. Furthermore, your opponent can make mistakes.
With these 2 concepts being the main backdrop, your goal or strategy should be clear. You need to really get 'in the lab' and think of ways to minimise risk, maximise reward and and take your opponent's game plan out of it. Let me use a really geeky real life example. Just skip this next paragraph if you don't want a major geek off.
I play pokemon, and this game is threat, counter-threat. I can bring in a threat and you will respond by bringing in your planned counter to it. My plan usually is to find hard hitting things that OVERLOAD on the same counter. In other words, you need your counter to be healthy to take on both of my threats. I can sacrifice one of my threats if it means crippling your counter, because it means the other can can then sweep your team. The thing with this strategy is that it doesn't matter what my opponent does if I execute it right. That counter needs to come in or he's gonna get his shit pushed in, but if he does, he'll lose the game either way. Win/win.
Of course, this is easier said than done, but that's why you spend so much time thinking in the lab, looking for strategies and combinations like this. The best way to play any game imo is to dig for elaborate ways to win the game without making a big risk. Find a
Xanatos Gambit or something close to it, as close as you can get. Find ways to push forward even if your plan A doesn't work.
Sorry for the rant, but I swear this answers your original question. To me, I've hit upon the key to doing well in any game, pretty much. My strategies will always revolve around this core idea. Why would I look for anything else? I can incorporate other ideas into it, in small ways, but this will always be the alpha and omega for me.