You can hopefully draw your own conclusions.
Sure, let's give that a quick stab.
The following should be taken as general commentary, not specific to the aforementioned service. Your mileage may (or rather, will) vary.
Here's the input data:
Source: The screenshots. I measured pixels over the height of the Y axis to get to preliminary numbers. Then for each of the two graphs I took the actually cited revenue to apply a correction to the figures to get to the likely actual numbers.
Note that this is an X/Y plot, as the screenshots' bar graphs use two different scales (1 bar per 8 hours vs 1 bar per 12 hours).The line graph is the accumulated revenue. I think it should be clear that it declines. While the sample set is too small (given variability) to really see it, this actually follows the changes in difficulty.
period | difficulty |
04/29/14 | 8000872136 |
05/12/14 | 8853416309 |
05/24/14 | 10455720138 |
06/05/14 | 11756551917 |
06/18/14 | 13462580115 |
06/29/14 | 16818461371 |
Source:
https://bitcoinwisdom.com/bitcoin/difficultyFor those unfamiliar, the generally accepted rule (aside from some shenanigans with pools' payout methods, for example) is that for a fixed hash rate, if the difficulty doubles, your revenue halves. Let's see if this holds somewhat by taking the relative difficulty change and multiplying the revenue from the previous period by the inverse of that number to get an estimated revenue.
period | actual revenue | 1/relative difficulty change | estimated revenue |
05/12/14-05/24/14 | 0.701 | n/a | n/a |
05/24/14-06/05/14 | 0.570 | 0.847 | 0.593 |
06/05/14-06/18/14 | 0.504 | 0.889 | 0.507 |
06/18/14-06/29/14 | 0.419 | 0.873 | 0.44 |
Well, that looks about right. The estimates are a bit optimistic but, given variability and all that jazz, are reasonable.
That means that we can do projections. Now the historic difficulty shows us that the difficulty rarely went below 10%, so let's include at least that one. Let's also include the ubiquitously cited 20%.
Note that difficulty changes should ideally happen every 14 days. However, if the network is actually 10% 'fast', then difficulty changes happen every 14/1.1 ~= 12.73 days. Similarly, 20% results in ~ 11.67 days. Which is why the dots on the two extrapolations don't line up vertically.This is what the two difficulty graphs would look like through to 2015/01/01:
Now, you can probably guess what happens to the revenue in that time period. Note that the 'historic revenue' ends at the last difficulty change.
At 20% difficulty increases, the sum revenue already starts to flatten out. At 10% difficulty increases, it's not so bad. Let's extend the time period through to, oh I don't know, 2020-01-01.
Note that there's an almost imperceptible discontinuity in the chart - this is because some time in 2016, the reward will halve. The block height at the last difficulty change was block 308,448.
Source: https://blockchain.info/block/00000000000000000828e7f7f643a00d1449041dfff89775139ab8e5ed2d7d58
Each difficulty adjustment period is 2,016 blocks, and the next reward change happens at block 420,000, so we can easily adjust revenue accordingly.
Source: https://en.bitcoin.it/wiki/Controlled_supply#Projected_Bitcoins_Short_Term
However, revenue per period at that point is already so low, that you can barely see it, even in the 10% graph.Well, by that time they have both well and truly hit their ceiling. Just to note, the numbers are:
For a 10% difficulty increase, total revenue:
BTC6.834
For a 20% difficulty increase, total revenue:
BTC4.759
That doesn't sound so bad, right? But then we get to that other number in the screenshot:
Purchase Price: $4,154.
The start date was 2014-04-04.
The USD/BTC at that time was ~$435
Source:
https://blockchain.info/charts/market-priceSo for $4,154 , one could have purchased $4,154/$435 =
BTC9.549.
If the difficulty only increased by ~5.83% (or less), then this particular example would exceed that figure by the time it ceilings out (If setting a cut-off of revenue-per-period of BTC0.001: mid-August 2016)Your mileage will vary. If you purchase BTC when it's relatively high, cloud hashing may be more attractive. While if you buy low, you'd be much better off just purchasing directly. But then cloud hashing companies are likely to adjust their prices, too.
And if you're thinking "well, I don't care about ROI in BTC - The BTC has been spent, it's gone, now I just want to know if I can still get ahead in fiat."
With the 10% increase, USD/BTC would have to be ~$608 at the ceiling to have break-even.
With the 20% increase, USD/BTC would have to be ~$873 at the ceiling to have break-even.
USD/BTC has been greater than those before, so there is that.
But just to rub it in anyway, assume those USD/BTC rates if you had purchased the BTC instead:
BTC9.549 * $608/BTC = $5,805 ($1,650 profit)
BTC9.549 * $873/BTC = $8,335 ($4,181 profit)
Disclaimer: This is just a quick stab at things - if there's a grave error, please do point it out (try not to quote the entire post, though - it's lengthy enough as is).