In this work we propose to use Twitter to find companies with a good growth potential that could be good investment options. In order to achieve this we built a sentiment model using the text content of tweets. We make use of hashtags to collect Twitter posts from a broad range of emotions so that our sentiment model can reliably distinguish tweets containing different sentiment expressions. To guarantee that no human sentiment is left behind we adopted emotions from the Circumplex Model of Affect and their synonyms and used them as search terms on the Twitter API. Afterwards, we use those tweets with a support vector machine (SVM) classifier to build a sentiment model. This model was used to classify company related tweets in order to predict the predominant sentiment in them. With the sentiment measures of tweets from different companies we created a trading rule that was optimized by a Genetic Algorithm (GA) so that we can maximize profit. Our simulations show that using the rules we implemented it is possible to build a profitable strategy for trading in the stock market using Twitter with the rules we implemented. During our testing period (November 7, 2016 to December 16, 2016) we achieved a 11% return, outperforming the SandP 500, NASDAQ 100 and DJIA composites.
