design.txt

Design document for stockx

For each stock, do this:
Get latest news for the stock symbol

Training neural net:
Run news through sentiment analyzer, which returns objective values for
positivity, neutrality and negativity
Use these as attributes to generate probability that the stock will have a
positive trend
Use next day stock values as results
Split data into .66 and .33, for training and testing respectively

Using neural net:
Run news through sentiment analyzer, which returns objective values for
positivity, neutrality and negativity
Use these as attributes for the neural net to generate probability that the
stock will have a positive trend

Tasks:
Find/create an API that returns latest stock news given stock symbol
Sentiment analyzer that returns objective values for positivity, neutrality and
negativity for some given text
Neural network that takes in the sentiment analyzer output as attributes, and
returns predictions for positive trends in all stocks

Notes:
Historical data:
http://ichart.finance.yahoo.com/table.csv?s={symbol}&a={month}&b={day}&c={year}
News:
http://www.reuters.com/finance/stocks/companyNews?symbol=AAPL.O&date=01082014
http://feeds.finance.yahoo.com/rss/2.0/headline?s=a&region=US&lang=en-US