StockBae

• 

  output json Screen shot

Inspiration

It's part of the huge product I'm building. With various sources and ways to analyze stock market , the technology today helps us analyze news , tweets , social media sentiments towards them. Though there exist lot of traditional algorithms , it's high time we start respecting the sentiments of the investors/clients have towards a stock. This is highly influenced by News / Media.

One mans loss could be other mans gain. That's one of the key concepts I've seen over and over again in the stock / financial investment Industry. A news , which could be POSITIVE for many , could have a negative impact for a investor. It's highly possible a NEGATIVE emotion towards a company could be leveraged to gain profit for our clients.

What it does

It analysis the latest news about the given stock. The API used here is Guardian API to fetch the latest news. It uses this article to calculate 2 things :

1.Polarization : how strongly polarized this article is. It doesn't matter if it's positive or negative. This is to intimate the client that , that stock is something he should analyze and make his own decision. A highly polarized stock , means there is high chance he could leverage the situation.

2.Sentiment: this is an indication to the client , another information , to help him decide what sort of a decision he would want to make.

How I built it

1. Did market analysis on existing stock market technologies available
2. Did Technical research on possibility of the turning the idea into working model
3. Decided on API - Yahoo-Finance , Guardian News , Algorithms - Naive Bayes
4. Tested each part of the code separately
5. Integrated the product so that the output of Yahoo-finance is fed into Guardian News , whose article is fed into the algorithms to get polarity and sentiment
6. Created web services for integrating services
7. Reviewed and Submitted B-)

Challenges I ran into

Though I initially planned the project based on IBM Alchemy News / IBM Alchemy API , EventRegistry.org news API , they all caused trouble after 3-4 hours , as it restricted the number of possible query per day. So this means , I had to rewrite all of it and decide on a different API.

After doing some research , I settled on Guardian API for news.

Since IBM Alchemy API couldn't be used for sentiment analysis - we had to use our own machine learning algorithms for it. We did our research on the various ML algorithms available , then we settled on one that's easier to implement in the given time and also widely used. Then with the test data already available , we tested it with example articles to ensure it's usable.

Accomplishments that I'm proud of

1.Ability to quickly make decision based on availability 2.Ablity to learn quick and implement solutions 3.Building Teams 4.Effiency in splitting task so that it could be worked on individually

What I learned

All great journey starts with one small step. That is to start building it and work on turning ideas into reality step by step.

What's next for StockBae

There is a looooooong way to go. This is just the beginning of the StockBae. This is one of the many ways we can predict / gauge a stock would go up or down. This is just one of the many services StockBae is gonna use. In future implementation , we shall start incorporating analysis other social media inputs. We shall also include other traditional stock analysis algorithm. Train the application with all these data to come to a more efficient way of helping clients gauge a possible situation which could be leveraged for high profits and cash outs.

Built With

• flask
• guardian
• machine-learning
• naivebayes
• python
• yahoo-finance