No, I put the green line on myself. It seems at least as good a fit to the data as the cyclic one, and I doodled it in 10 seconds.
My point is that with such a small amount of data, you can't just plonk a regression curve onto it and call it a day; you can get any number of different curves of different kinds to make a good fit. Linear, quadratic, some fourier terms... but your choice can be arbitrary and highly leading and suggestive. The one you provided made it appear cyclic. Mine made it appear a distinct upwards trend.
In short, that data is too ambiguous to analyse like that.