Stocks/investments are always so confusing - people are easily inundated with an overload of information/numbers. With InvestBit, we aim to make complex concepts and ideas into easily digestible bits (or bytes) of information.
How it works
There are two parts - equities analysis and stock analyst analysis. This is all hosted and stored through IBM's BlueMix as a web application. MatLab and Python were employed to scrape an external website [wealthminder.com] and download extra analyst data; this data was then processed and analyzed along with SEC data provided through Finra in Matlab. PHP was primarily used along with Yahoo's free finance API to pull up various financial information about stocks in real time.
Challenges I ran into
IBM's BlueMix had a very experimental SQL module; as a result the database we were supposed to use to store analyst information did not work (we couldn't connect to it through an external port). A quick, dirty (and wildly inefficient) workaround we had was to export the entire table as a .csv (comma separated value) file - that way the data can still be relatively easily read and processed. The analyst data provided through Finra was very...rudimentary. There was no information in FInra's database that we felt was actually useful in determining an analyst's success, as a result we wrote a Python script to parse through wealthminder.com as it had a list of analysts as well as their portfolio size. We felt that portfolio size (how much money an analyst is managing) is a good indicator for success - we assumed that the more money an analyst is managing, the more trustworthy/dependable the analyst should be.
Accomplishments that I'm proud of
-I never actually wrote such a complete Python webscraping script - the script literally ran through the entire website and downloaded information on every analyst recursively. -Getting, combining, analyzing, processing and actually understanding the data we had through MatLab was a huge accomplishment - we dealt with humongous (tens of thousands) of analysts, and had to find a way to efficiently find relationships between the various data sources that we had. In addition, all of this data is publicly available - and that's what really amazed us. It wasn't easy to get to the data (the webscraping was not fun) but anyone can do this and run data analysis as long as they have a laptop or computer.
What I learned
-We all learned from each other, as each team member brought specific skills and technical expertise in a different area. For example, Brian Chen already had prior knowledge on financial reports, so having him on to build the equities analysis page was a no-brainer. Nick Walsh had prior experience in data analysis in MatLab, so everyone provided him with the information and he showed us how he figures out relationships. Brian Yang is the resident CSS wizard and web administrator, so he was responsible for setting up the website on IBM's BlueMix and having all the connections between the various modules of the project mesh together both technically and visually.
What's next for InvestBit
-Expanding the algorithm/taking other approaches to analyze both analysts and stocks. We hope to have a predictive algorithm to both predict an analyst's effectiveness and a stock's profitability after enough time/development on the project!