John from howigrowmyweath.com wrote a great article on Ethereum’s cloud miming back in April 2017. Although hashrate and price has varied a lot since then, basic ideas are the same.
Firstly, in his analysis he assumes price to remain the same over the period of mining (~48$).
If you buy 1 MH/s for 22$ from a cloud mining service and start mining with a hashrate of 22595 GH/s, then you are going to earn 0.043 ETH per month or (~3 dollars) in a year. About a 13% margin rate. Seems not bad.
But here is what happens if the hashrate rises (and it does). Assuming a 1000% hashrate increase (reasonable guess according to history) the volume of ETH mined per month decreases to 0.004. Taking into consideration a slight exponential decline in ETH mined, the overall ETH mined during the year will be 0.28. That means around 13.8$ per year (assuming 1 ETH~48$). Hence, we end up losing around 8 dollars (or 36%) of our initial investment.
The author also shows that the results of mining are not positive even if the growth rate of hashrate does not grow so sharply (with a level of 500% or 100%).
The author concludes that cloud mining is not profitable unless you get serious (50-70%) discount on mining power.