This article uses Tweets from social media platform, Twitter, as ground truth for generating buy/sell signals. This algorithm was used on different stocks to test the potential for different language use when talking about a stock. Overall the results were somewhat processing, but for better performance, specifically tailored dictionaries would be more beneficial.
Key Points
Using stock specific dictionaries can help better classify tweets
Low tweet rate for some companies leads data to be inaccurate
Citation
@INPROCEEDINGS{8628841, author={Birbeck, Ellie and Cliff, Dave}, booktitle={2018 IEEE Symposium Series on Computational Intelligence (SSCI)}, title={Using Stock Prices as Ground Truth in Sentiment Analysis to Generate Profitable Trading Signals}, year={2018}, volume={}, number={}, pages={1868-1875}, doi={10.1109/SSCI.2018.8628841}}
Title
Using Stock Prices as Ground Truth in Sentiment Analysis to Generate Profitable Trading Signals
URL
https://arxiv.org/pdf/1811.02886.pdf
Summary
This article uses Tweets from social media platform, Twitter, as ground truth for generating buy/sell signals. This algorithm was used on different stocks to test the potential for different language use when talking about a stock. Overall the results were somewhat processing, but for better performance, specifically tailored dictionaries would be more beneficial.
Key Points
Citation
@INPROCEEDINGS{8628841, author={Birbeck, Ellie and Cliff, Dave}, booktitle={2018 IEEE Symposium Series on Computational Intelligence (SSCI)}, title={Using Stock Prices as Ground Truth in Sentiment Analysis to Generate Profitable Trading Signals}, year={2018}, volume={}, number={}, pages={1868-1875}, doi={10.1109/SSCI.2018.8628841}}
Repo link