ETH Mining - Your profitability after two years (Using statistic models to solve the question)
Anyone who follow my story would know I will give a daily update on my 10mh ETH contract in Genesis Mining. The purpose is to help everyone to make an informed decision before joining the program.
But even with the daily update, I feel there are still an unanswered question as a lot of people will say there won't be any profit considering the difficulty pick up while there are another group of people saying vice versa.
I know the ETH may be going to Proof of Stake soon, but nonetheless this model could fit in to other alt coins mining as well, So why not?
My goal in this article is to provide another perspective for more meaningful discussion and I feel by using some simple statistic models would do the job.
The first step I will need to do is to reconcile the number in the profitability calculator :
Seconds per month: 2,592,000
Total Network hashpower: 62510 GH/s
My hash power: 10mh/s
Block Time : 15 sec
Each block reward : 5ETH
So ETH per month would be equal to = (10 / (62,510* 1000) * (2,592,000 / 15))*5 = 0.1382 ETH / month
OK, this is a good start.
So now, you can see the only variable I need is the Network hash rate which is dependent on the mining profitability and others. But I want to make it as easy as possible and the only variable I will use to forecast the future network hash rate is Time. With that, the second step is to develop a statistic model such that I can extrapolate the data and forecast the future.
Geometric Mean Model (The not so good model)
By using the data provided in below website, referencing the past 1 year data (from 2016 July to 2017 July), you may easily calculate the daily growth rate of the network hash rate to be around 0.74%. With that, I can project the network hash rate in 2 years.
https://etherscan.io/chart/hashrate
As shown in below chart, the hash rate will go up to 12,652,320 GH/s which would be increased by around 20792% from today. If you ask me, this may not be reasonable given that it would not be economical to mine and so this will make the difficulty stablize and there are simply not that many GPUs to be provided in the market.
A simple regression model
The idea is to draw one single straight line that would fit the current data the best and therefore extrapolating the future using that predicted line. There is a statistic terms on that (but for the purpose of this analysis, please assume that the smaller squared error is, the better the model is).
Now this looks a bit more reasonable and it improves against the previous model (in terms of squared error). This model predicts the hash rate will be increased to around 106,549 GH/s in two years which represents 76% increase from today.
However, as you may wonder the model did not pick up the recent spike in ETH difficulty very well and this leads us to consider the third model.
Exponential regression model
Because the ETH difficulty do not increase linearly and therefore it require some other statistic models that would fit the line better and an exponential regression model will do the job.This time you will see the tail of the estimated line (in orange) is trending up as well which should give better fit to the data and therefore better projection to the future.
So with all the maths, what does that mean?
If we are able to estimate the future network hash rate, it means we can use the simple formula (as discussed in the earlier of this post) to calculate how many ETH we can mine everyday and therefore reach our goal to calculate the potential profitability of our 2 years contract. I won’t explain too much of the statistic outputs and you may comment in the post to let me know how you think about it :).
Statistically, the Exponential method will give the best estimate like I discussed before.
Estimated ETH Network hash rate in two years.
Finally, the predicted profitability of our Genesis Mining contract.
If you want to see similar analysis on other alt coins, let me know in the post :)