The recent US Election has been a roller coaster ride. Perceptions towards candidates continuously fluctuated following every interview, debate and allegation. We have gathered some information from our friends at RavenPack and Amareos to show how sentiment analysis predicted the election.
Through this election as well as Brexit voting, we have definitely learnt that polls aren’t as reliable as they used to be. What is coming to light though, is the definite growing importance of sentiment analysis. Sentiment Analysis aggregates data from many media platforms (editorial news, commentary and social media all included) and depicts the thought processes of its authors. In these two historical political moments, sentiment analysis has proven to be more accurate than polls. Amareos predicted Trumps victory based on crowd sourced data from twitter sentiment.
3- Day rolling Average for Sentiment Spread vs polls spread
RavenPack conducted a media sentiment study of the elections for a 14- day and a 3- day rolling average for sentiment spread vs polls spread. The poll numbers have been extracted from the Financial Times and RavenPack has calculated average sentiment levels on a daily basis across various categories for each candidate to deduce the sentiment spread. Numerous ups and downs have been seen during these elections and those evidently impact the public opinion of the candidate as seen in the graph above. Clinton was leading before Director Comey reopened the investigation on Mrs. Clinton on October 28th. And then on November 6th, once the FBI cleared her of all charges, the sentiment was leaning in her favor again as indicated by the graph above.
Now that the elections have come to an end, Trumps victory has brought uncertainty to the Global Financial Markets. This is simply because we don’t know what kind of president Trump will be. Will he be the erratic person who threatened to lock up his political rivals during the campaign? Or is he capable of being a rational leader of the free world?