Since I made that chart, i'll say use it with caution. If you make any trading decisions based on that chart you are on your own. It only contains data for part of the period, after the first large bubble in 2011. It also has to be expanded with a S-kind of ending at the top since it will obviously flatten out when saturation is reached. But it gives the best fit to the data for that time period -- under the assumption that a single function generated the data during the entire period.
Fixed

Not attacking you, by the way, I appreciate your input. Just that it's a pet peeve of mine: pointing out that despite all the other assumptions (remove outliers or not?, which period to use for input?, exponential function? double exponential?), the assumption that it is exactly one function we're looking for is perhaps the biggest (simplifying) assumption of them all.
Yes, I agree. It's the best "simple" function I have found that best represent the data in this period. I'm sure there is more to it than this. Maybe step functions and