One of the more enjoyable elements of learning to program in both R and python has been the potential value extracted from big data. Of course, this is a buzzword but it also has the scope to provide potentially valuable on the ground insights.
The first example comes from our friends at sentieo – who have analyzed meta-data from Alexa to predict revenues at Twitter, as shown below:
https://app.sentieo.com/api/chart_viewer/?p_id=5a7a23983839814d9d0002e2
One of the useful element of this is that Sentieo has taken readily available data to translate into some potentially meaningful revenue forecast other than pure guesswork.
Of course, sceptics will argue that this is not reflective of all investment cases and indeed in other sectors, a more granular approach is necessary.
I certainly agree – one of the elements I have been picking up through programming is the extraction of readily available micro-data and translating this into investment themes. Of course, this can be quite hard as many data providers may use pdf formats and translating this into a CSV is a challenge unto itself.
However, the implications are profound. I am in the process of developing a micro-data set which could provide not only meaningful extraction of data but also the potential to see predictive elements through machine learning. Ultimately, I think there is scope for developing independent patterns of revenue recognition based on micro-data that can give a more meaningful outlook on revenues other than the thought process of ‘consensus is too bearish.’
This perhaps may, in a vacuum, seem vague, and I will provide further examples as we go along and explore this topic. I am a programmer in my spare time so I promise it gets fun!
Thanks, Rickin
Rickin Thakrar - my site 