What I meant was that other risks do not justify a higher offered ROI. Exchange rate is a risk but not relevant to the pricing of the product due to the law of one price (spot-future parity).
Thats incorrect. You're ignoring maintenance fees which are function of exchange rate since electricity, hosting and usually hardware is still priced in fiat. Maintenance fees currently take up 50-90+% of the payouts on most legitimate contracts, so this is pretty darn crucial. If BTC falls below $200, all these contracts become worthless. If BTC goes ballistic, even cex.io contracts might make you a nice profit. AT least until difficulty catches up.
That leaves future difficulty (expected mining income) as the only factor that determines the price, which must result in zero expected ROI at inception for similar reasons. Buying 250 LTC (expected) for 300 LTC is obviously a bad deal, while the other way around is bad business (plus it would create an arbitrage opportunity). So the price must equal the expected coins mined with the contract, so that the NPV equals zero (or at least head in that direction).
Again, you're assuming to know how much a contract will pay out. If you do not know what difficulty will be like 6 months from now and what the exchange rate will be, how do you know what the expected payout will be ?
You see cloudmining as a loan from the investor to the miner, I see it as the miner selling off a huge risk to the highest bidder.