On any given day, it has almost always been the case that simply buying Ethereum tokens (ETH) would have been more profitable than mining it. For example, even if you started on day 1 (July 29, 2015) at home, you'd have half as many ethers as someone who bought on an exchange -- or somewhat less than someone who bought on an exchange any time up to January 21st, 2016 (give or take). A miner starting on the release day of the RX 480 (June 29, 2016), paying MSRP, would have as many ethers as someone buying ethers on the open market any time up to January 23, 2017.
Conditions required for mining to be more profitable are that difficulty remains low for a prolonged period of time before a rapid rise in price since rising prices can be expected to result in rising difficulty (but the opposite is not true -- if the price increases by 10% and the collective hash rate increases by more than 10% to take advantage of the price increase, this will not drag up the price). Your "investment" in hardware is depreciating in the meantime and your capital is tied up. Alternately, miners come out ahead if the price of Ethereum plunges, in which case the miners can still sell off/use physical assets.
I wrote a program to use historical data to determine whether one would have more ETH buy simply buying and holding ethers or by mining them. There are several caveats in the calculations (see program code), but regardless of how you calculate it, it's still true that mining has rarely resulted in better yields than buying ether. The program also has the ability to extrapolate (it defaults to doing no extrapolation).
Past performance is not indicative of future returns and this answer, nor the program, constitute investment advice. These are provided merely as informative tools (with no guarantees of fitness for purpose) and are provided as-is.