Dynamic and Context-Dependent Stock Price Prediction Using Attention Modules and News Sentiment

Abstract: 

The growth of machine-readable data in finance, such as alternative data, requires new modeling techniques that can handle non-stationary and non-parametric data. Further, non-stationary time series data is common in finance. In the case of price series, each value depends on a long history of prior levels. However, most machine learning models that predict financial time series expect stationary inputs. Consequently, these models rely on standard stationarity transformations, such as integer differentiation, to produce returns, which have a memory cut-off hence the series loses important signals. In addition, in the field of causal inference, the concept of independent or autonomous processes has proven to be important because these processes make the model capable of cause and effect inference. Thus, a complex model may be thought of as a collection of separate processes or “causal” modules. As a result, individual modules may be robust or invariant even when other modules change, as in a distribution shift, and therefore may have a better overall generalization.

This talk will look in depth at the use of a new modeling approach, the alpha-t RIM (recurrent independent mechanism) for industrial time series prediction. The presented architecture makes use of key-value attention to integrate top-down and bottom-up information in a context-dependent and dynamic way. You will learn how to model the data in such a dynamic manner, with the use the alpha-t RIM, which utilizes an exponentially smoothed recurrent neural network combined with a modular and independent recurrent structure. As a practical use case, we will look how to forecast stock prices from the S&P 500 universe with the use of their news sentiment score and historical pricing data.

Background Knowledge
Intermediate - Advanced, though they should be familiar with deep learning, I will explain the background.

Bio: 

Nicole is a Data Scientist & Quant and Data Engineer currently working at impactvise as Data Science and Technology Lead and at quantmate as Quant. She has over 8 years of experience leading technology projects. She additionally reviews machine learning books and online courses for Manning Publications. Her research interests include time series prediction and natural language processing. She is dedicated to showing others how to succeed in machine learning and is committed to making STEM more attractive to women.

Open Data Science

 

 

 

Open Data Science
One Broadway
Cambridge, MA 02142
info@odsc.com

Privacy Settings
We use cookies to enhance your experience while using our website. If you are using our Services via a browser you can restrict, block or remove cookies through your web browser settings. We also use content and scripts from third parties that may use tracking technologies. You can selectively provide your consent below to allow such third party embeds. For complete information about the cookies we use, data we collect and how we process them, please check our Privacy Policy
Youtube
Consent to display content from - Youtube
Vimeo
Consent to display content from - Vimeo
Google Maps
Consent to display content from - Google