Predicting stock price movements based on newspaper articles using a novel deep learning approach

WOE 16:00 - 16:30

Lezingenzaal 8

QUESTION: What was worth millions of dollars yesterday, but it is not worth anything tomorrow?

ANSWER: Today's newspaper.

For my master thesis of the Business Analytics program at the Vrije Universiteit, Amsterdam, commissioned by Deloitte, I developed an algorithm that learned to predict stock price movements, based on news paper articles. It took over twenty six thousand articles to train the model, but in the end it was capable of predicting over 56% of articles correctly.

This presentation will walk through the algorithm's pipeline from begin to end in an intuitive, yet technical matter. After they have left the presentation, visitors are expected to have a global understanding of the components of the algorithm as well as why and how they work. The components discusses include: Word2Vec, Long-Short Term Memory (LSTM) and a Convolutional Neural Network.

  • Thema
    Text Mining

    Text mining

    Bezoekersdoelgroep
    Data applicaties ontwikkelen

    Deze lezing gaat concreet in op de techniek en het gebruik van data. Er wordt uitgelegd hoe bepaalde dingen aangepakt en wat daarvoor gebruikt is. Hierbij gaat de spreker de diepte in en geeft deze specifieke feiten, inzichten, advies en tactieken op hoog niveau.

    Deze lezing is bedoeld voor bezoekers die: data applicaties ontwikkelen, zoals o.a.; data engineers en data scientists.