Event_driven_stock_prediction

This project is about event-driven stock prediction. It analyzes the impact of news to the stock price movement.

Model Development

Basic Model: Neural Network with 2 hidden layers. 1st hidden layer: 300 nodes, tanh 2nd hidden layer: 50 nodes, tanh output layer: 1 node, sigmoid

Version 1: S&P500, NYSE Volume, Nasdaq Composite, 5-day lag time series for the stock

Version 2: Version 1 + Sentiment Polarity for news

Version 3: Version 1 + average word embeddings for the most confidencial sentence(filter by ReVerb, http://reverb.cs.washington.edu/)

Version 4: Version 1 + Doc2Vec(all news content)

News actually can improve the performance of stock predictions. But it is hard to decide the best model based on the facts below:

Xu Gao: the leader, initial research, news parsing, model design

Scott Edenbaum: SQL database created and BeautifulSoup scraping

Name		Name	Last commit message	Last commit date
Latest commit History 69 Commits
config		config
dev		dev
news		news
notes		notes
src		src
tests		tests
.DS_Store		.DS_Store
.gitignore		.gitignore
.php_cs		.php_cs
.travis.yml		.travis.yml
1dCnn.py		1dCnn.py
BAC.py		BAC.py
LICENSE.txt		LICENSE.txt
NN.py		NN.py
NN.pyc		NN.pyc
NN_no.py		NN_no.py
README.md		README.md
Scrape30.py		Scrape30.py
cnn.py		cnn.py
composer.json		composer.json
dbgrab.py		dbgrab.py
dbgrab.pyc		dbgrab.pyc
doc2vec.py		doc2vec.py
importcommon.py		importcommon.py
importdb.py		importdb.py
market.py		market.py
new.txt		new.txt
output.txt		output.txt
parse.py		parse.py
parse_reverb.ipynb		parse_reverb.ipynb
phpunit.xml		phpunit.xml
pricedb.sql		pricedb.sql
result.xlsx		result.xlsx
reverb-multi.py		reverb-multi.py
reverb.py		reverb.py
seekingAlphaScrape.py		seekingAlphaScrape.py
seekingAlphaScrape300.py		seekingAlphaScrape300.py
sentiment.py		sentiment.py
sentiment.pyc		sentiment.pyc
single_news.txt		single_news.txt
table.csv		table.csv
tickerScrape.py		tickerScrape.py
tickerScrape1.py		tickerScrape1.py
tickerlist		tickerlist
ts-input.py		ts-input.py
word2vec.py		word2vec.py