.Olai Project

Share price forecasting tool for use with companies on the Oslo Stock Exchange (“.OL”) with artificial intelligence (“AI”) tools

Proposal


A customised lexicon per company and/or sector, could improve sentiment analysis due to a greater emphasis on the revealed preferences and behaviours of investors for a range of news items.


Hypothesis

There are inherent short-comings from applying an abstract model to specific circumstances. This problem could arise when sentiment analysis is applied to stock market news.


Problem under consideration 

Sentiment analysis often relies on tokenisation to prescribe value to individual words, conducted independently to the specific circumstance. These values lack market context and the preferences of investors.


Validation test on 2019 data

Introductory slides