analysis of the feelings of internet content and especially social media is a technique used more and more in the e-Reputation. Recently, it is also possible to analyse images and especially video streams to infer the State of mind of a person. The french “Start me up” offers of such algorithms of image analysis as well as the American Emotient, bought by Apple earlier this year. Merchandising, marketing and advertising are the first targeted, but a professor in finance saw unexpected use.
5,000 microexpressions spied to assess the feelings of the interviewed
Fear, anger, disgust or surprise, this is the kind of feelings that an algorithm is able to extract a video stream, such as a CEO who defends his company’s quarterly results. This data, James Cicon Professor in finance from the University of Missouri it was correlated with his company’s financial data: margins, return on investment, movement of course, etc.
Dr. Cicon rotates a parsing algorithm of feelings about 200 videos of CEO of Fortune 500 company and the correlation was found to be: a CEO that reflects a certain disgust in an interview in fact hides an average profit increase of 9.3% in the following quarter. On the contrary, a CEO who blurts out signs of fear is harbinger of an increase in the action in the coming weeks. Results paradoxical but that psychologists explain. Fear is regarded as a strong motivation for the CEO.
The algorithm developed by Dr. Cicon relies on the work of Dr. Paul Ekman. In the 1970s, it defined a catalogue of 5,000 microexpressions and, according to Professor of the University of Missouri, the algorithm would have a reliability of 90%. A result especially if it is proven, what drives the Professor of finance to consider the next step: evaluate the sincerity and truthfulness of the speech of the CEO. If this new variable enriches trading algorithms, it is Hollywood that companies will have to recruit their CEO
Source: ‘Software Detects CEO Emotions, Predicts Financial Performance’, 17 February 2016