Ok, let's try again. This machine costs 3.5 BTC. It returns 2.5 BTC in a year. You lose 1 BTC.
I can just buy 3.5 BTC and then not buy the miner. In a year I still have 3.5 BTC. No matter what the cost of BTC is I have more $$ than if I bought the miner.
Yes, BTC price increase will affect your fiat ROI. But look above. If you can't ROI in BTC you would have done better to hold the coin. If BTC goes up, you have more fiat if you hold the coin. If BTC goes down you lose less money if you hold the coin. Mining at a BTC loss is willingly increasing the risk to your fiat while simultaneously diminishing the returns of your fiat. You make less for higher risk.
Would you buy 4 gold bars to trade for digging equipment designed to dig up 3 gold bars and then hope for the price of gold to go up? No you wouldn't, you would just hold the 4 gold bars. Why should BTC be any different? Why would you pay more BTC than a machine returns?
This is why you run ROI in BTC. If you can't recoup BTC, hold the coin. If you can recoup coin plus more, decide if the 'plus more' is worth the risk (will the machine and variables hold up).
Personally, this just seems like way too long of a ROI for bitcoin mining gear. I get free electricity and I look at gear that i can pay off within 60-90 days.
i have free electricity as well but i wouldn't say roi needs to be 60-90 days. that's just giving customers money. a reasonable time frame to aim for is 6 months or less. what should have stuck out to s9 buyers was how bitmain sold a miner that based on current calculations would break even in way under 6 months. They never do that. but people were blinded by greed and got alot of money taken from them this round. hopefully they learn their lesson this time and dont buy new hardware that costs so fricken much! i wonder what the cost of the miner will be after halving. either way i know it will be a better roi if you wait for a later and probably less buggy batch.