Hey friends.
A little bit of background:
I spent last Spring conducting my undergrad thesis with the goal of developing a system which could predict stock price movements using data like social media posts and news articles. A few of you might remember my post in here way back in the day where I demoed a tool which allowed for some user interaction with the data that my system was collecting. To make a long story short, it became clear to me that introducing a new, complicated tool wasn't exactly what the average investor needed. This realization made me take a step back and keep more true to my mission: to package and deliver advanced data analytics to the average investor.
I've abandoned the tool approach and decided to pursue a newsletter system of sorts -- effectively, Quikfo will take in tons and tons of market data, analyze it in order to derive conclusions, package these derivations into a digestible format, and deliver this comprehensible information to a user's inbox. At the moment Quikfo is in BETA and the functionality is constantly expanding.
The meat and potatoes:
So I've put together a BETA on my site https://quikfo.com -- please feel free to check it out.
As 99% of my attention has been spent on backend functionality, such as data collection, analysis of said data, and packaging of said data, the frontend and user pages are mediocre at best. Eventually I'll get around to fleshing out more of the user interface and add profile pages and whatnot, but for the time being it's just not at the top of the todo list.
At this point in time as a Quikfo user you can subscribe to a single formal data package entitled 'Bullish Sentiment Indicators'. Quikfo contains a microservice which leverages Bayesian classification (a machine-learning technique) in order to determine the relative 'bullishness' of words; this microservice can determine whether a company's current diction among news and social media matches the diction of companies who have recently seen abnormally bullish trends. In the not too distant future I expect to have formal packages for all sorts of things like 'companies seeing positive social media PR', 'companies seeing negative social media PR', 'companies seeing positive news PR', 'companies seeing negative news PR', 'companies with big PR swings recently', 'companies being talked about more than normal', 'companies being upgraded by analysts', 'companies being downgraded by analysts', etc. as well as packages for things like oil futures, gold futures, iPhone sale analysis, etc.. The possibilities are really endless, it's just a matter of getting things done.
Users can also enter in custom interests at the moment which will populate their newsletters with less cleanly packaged content, but hopefully relevant content nonetheless. For example, since there's no formal package for monitoring clinical trials at the moment, maybe you would want to enter 'trials' as a custom interest. This should provide you with, at the very least, news articles and social media posts delivered to your inbox each day which could contain useful insights. This feature is especially useful for me as a developer because it allows me to see what the most popular interests are which aren't currently formally offered, effectively telling me what I should be working to offer ASAP.
I'm in the process of partnering with my college to receive access to significant server infrastructure and databases capable of petascale processing which should really expand what I'm able to produce content wise.
The gist of it:
I built https://quikfo.com which you can sign up for in order to receive potentially useful insights on news and social media data -- it's early in development so please report any bugs and maintain high levels of patience.
Please feel free to ask me any questions, I'll get back to you in the comments as soon as I can!
Submitted October 03, 2018 at 03:31PM by TheLoneDonut https://ift.tt/2O17zs0
No comments:
Post a Comment