When buying a bitcoin miner, your first concern is will you ROI on your purchase.
To know whether you will ROI, you need a good idea of how quickly the difficulty in the future will increase.
I've been working on making difficulty projections, and most of the time, no matter what forecast model I use, I project that I
will not ROI.
I've tried plain old exponential, double and triple exponential smoothings along with neural networks, double moving averages, polynomial, etc.
Here is a plot of the Bitcoin Difficulty over the past few years:
Note the long period between 2011 and 2013 that Bitcoin difficulty wasn't increasing very much due to ASICs not being on the market.
Let's zoom in on the ASIC era:
And now let's use double/triple exponential smoothing to forecast out 1 year:
Exponential smoothing was one of the more generous of the models I chose, and what I'm forecasting is that the projected difficulty increases wouldn't allow, for example, a SP31 that starts mining 10/20 to come even close to ROI'ing even with sub 5 cent electricity.
Here's the projected absolute difficulty:
Table of projected increases per jump:
8/31/2014 8.5%
9/12/2014 10.8%
9/24/2014 10.6%
10/6/2014 10.4%
10/18/2014 10.1%
10/30/2014 9.9%
11/11/2014 9.7%
11/23/2014 9.5%
12/5/2014 9.3%
12/17/2014 9.1%
12/29/2014 8.9%
1/10/2015 8.7%
1/22/2015 8.5%
2/3/2015 8.3%
2/15/2015 8.1%
2/27/2015 7.9%
3/11/2015 7.7%
3/23/2015 7.4%
4/4/2015 7.2%
4/16/2015 7.0%
4/28/2015 6.8%
5/10/2015 6.6%
5/22/2015 6.4%
6/3/2015 6.2%
6/15/2015 6.0%
6/27/2015 5.8%
7/9/2015 5.6%
7/21/2015 5.4%
8/2/2015 5.2%
8/14/2015 5.0%
8/26/2015 4.8%
Any thoughts anyone?
