I have looked at their spreadsheet and find it lacking. For one its based on 10day increments when they will be paying weekly. Second, their increase in network hashrate is fixed instead of an average %. I could go on.
Of course it's lacking. It's not easy to make an accurate model in spreadsheet.
Why 10 days?
They tried to follow difficulty update which happens on 10 to 14 days, depending on increase.
They are not calculating weakly divs, there is only 'total' divs/share for the whole year.
Why fixed increase?
Well, it's not fixed, look closer, it's changing, starting with 2500 (jan, feb), then it's 3500 (mar-may), after that 4000.
If you want that in % - that's something like starting with 20% increase now and decreasing that every retarget period by 7%. (maybe too optimistic?)
Of course its hard to make an accurate model. The spreadsheet is to try to make a prediction to help make investment decisions. I would use a % increase rather than a fixed # because we have historical data for the growth of the network. You could program a reduction in % after so many weeks, ie, 10% of 10%. this would account for the likely slowing of network growth over time.
I am also curious why they didn't just use the difficulty, instead of their % of the network.
I understand them using 10 days because of difficulty increase, but with weekly dividends and reinvestment, it would make more sense to calculate for the week, and perhaps calculate the difficulty increase weekly as well. While this wouldn't be exact, it would provide #s that are likely more conservative and things work out better than the math says.