Once you have coins in hand you know exactly what your break even point is - the price at which you bought in at plus exchange fees. People who bought ETH last year at $300 should have sensibly sold some (if not all) at $1200 to $1500. Once simple click would have resulted in 5x fiat profit, and should they like - they could always rebuy the ETH now and get 5x as much.
Of course this is all hindsight - but it does show the dichotomy - the strengths and weaknesses of both. If you had lots of capitol and were willing to take 100% loss people should have bought coins and not mining eq. If they wanted to just put 1 foot in the crypto ocean - they build rigs.
Here's hindsight - ETH had fallen all the way to $10, even $7 during last January when DASH starting during it's insane rise. Nobody wanted ETH (well except the believers - Polo trollbox chat showed this). If you bought then you would be up 30x from just holding 6 months longer than somebody buying in July. That's why it makes sense to look at the big picture.
ETH is the perfect of example of a coin that you shouldn't have mined (even though I did). If you bought the coin at launch is was sub 1 penny. You could never mine as much ETH with a single RX 480 going full tilt to today as if you had spent all the money on the coin at ICO.
So - if you expect the coin to fail mine it. If you expect it to succeed buy it (at least well timed buys - doesn't take a genius to not buy ETH at 1500 or BTC at 20K+).
Miners really should spend time reading the speculation and investing parts of this forum. Sticking your head into the mining forums 12 hours a day seeing if they can eek out 1 more sol out of 1060 by changing drivers or tweaking with settings are missing the forest through the trees.
Buying eth at ICO is the best case scenario. But if you missed the ICO? what then? A 200 $ rx480 bought in the summer of 2016 has mined a shitload of eth and on top of that it gives you other opportunities. In the end it comes down to individual preferences, some people mine other people trade.
OK, forget ICO. Let's pick Jan 1st 2017. You could have bought about 30 ETH for the price of a RX 480. Even with cheap 0.05cent power you still come out ahead by buying because ETH was successful. Its fiat price accelerated faster than miner could accumulate it.
You would need to look at a coin like LTC which mooned in 2014, crashed severely, then mooned again, then crashed again - there the fiat gains are less than what would one would have accrued from mining. But the whole ASIC thing throws monkey wrenches into the prediction.
Basically you would need to mine a coin that other people have given up on or had not taken interest in for you to come out ahead in the long run. In narrow windows like 3 months or 6 months it can go either way.