My alternative approach - log vs log(log) from 01.11.2011 to present. Conclusion: The log-log model seems to better fit the data than the linear-log model, and according to this model we are actually slightly under the trend line. Comments?