Where does this number of 15 days come from?
From the top of my memory I though ipollo mentioned an ROI in 3 months, assuming there are not enough miners around to increase the hash rate. Since I assume there are enough miners to increase the hash rate, an ROI in 6 months is more realistic I think. The biggest X factor is that it is uknown how many G1 mini miners have been produced and when the G1 normal miner will be released.
If you want to calculate ROI for the mini. Graph rate is 1.4, 100 W power consumption
So asuming for example electricity costs of 0.10, pool fee of 1% and 600 dollar for a G1 mini, the ROi would be in 91 days 
! Indeed amazing!
However, you should assume there will be enough miners to increase the hash rate, so expecting an ROI of 3-6 months is more realistic. In either case, it still is a good deal, hence I ordered two of these babies.
https://www.coincalculators.io/coin/grin_32?Hashrate=1.4&HashFactor=h&Watt=100&ElectricityPrice=0.1&PoolFee=1&HardwareCost=600&LH=&Difficulty=24
=> Another very nice part of the G1 miner, is the low power consumption of 100W, this means that even while mining on green electricity with a very high electricity price (e.g 0.27$), I still have a very good ROI. This is in general a nice and unique property of Grin’s Graph based proof of work algorithm, it is low on power consumption (good for the plannet), but requires good hardware/fast memory. This also means that even if the hash rate would increase drastically when more miners are released, running your G1 mini miner is still cost effective since electricity is not a big factor. So contrary to most miners, you can use them for many years and still mine grin cost effectively when much faster miners would be produced.