Sentiment Analysis and the related topic of Opinion Mining concerns the analysis of texts (news, social media, micro blogs) and turning these into sentiments, opinions even feelings and human emotions. This goes beyond simply analysing the number of likes, shares or comments we normally get after any marketing campaign and then reviewing it for the number of positives, negatives and measuring the level of success.
Sentiment Analysis is definitely creating waves in the finance sector. UNICOM organised on 18 – 19 June 2014, a very engaging Conference : Behavioural Models & Sentiment Analysis Applied to Finance. Major News Analytics vendors including Bloomberg, Thomson Reuters, RavenPack, OptiRisk Systems, Deltix, Northfield Information Services and MarketPsych participated and exhibited in the conference. The conference was based on four sections: Foundations & Technologies of Sentiment Analysis for Finance; Sentiment Analysis for Finance: Case Studies & Use Cases; Social Media, Micro Blogs, Google Trends; Sentiment Analysis for Multiple Asset Classes.
In this event, UNICOM brought together a unique group of industry researchers and academic whose work is exclusively focused in this domain. Over 145 professionals from the finance industry attended the conference.
Tobias Preis from Warwick Business School & Artemis Capital Asset Management Gmbh was an Opening Keynote. In his talk, he outlined some recent highlights of his research and presented “Quantifying Economic Behaviour Using Big Data”.
A research work on “Evolution of News Analytics” was presented by Gary Kazantsev, Head of R&D Machine Learning Group at Bloomberg. He highlighted Sentiment Analysis in broader way and presented better models for short texts (headlines, twitter etc) and Market Impact Indicators (Market Moving News).
Elijah DePalma, Senior Quantitative Research Analyst from Thomson Reuters introduced News Analytics and Text Analytics, Firm Level Sentiment, Market-wide Sentiment in his agenda of presentation. In his talk “Sentiment and Investors Behaviour” he shared Thomson Reuters News Archive data from 1996 – present.
Exploiting Entity Co-referencing in Unstructured News was explained by Peter Hafez, Director of Quantitative Research at RavenPack. Peter outlined Multi-dimensional predictive modelling and the impact of news across company “networks”.
Gautam Mitra, Managing Director of OptiRisk Systems showed results of News Stories: Measuring the impact on Asset Behaviour and Designing Trading Strategies. He first introduced a new metric “news impact” which combines: sentiment, decay of sentiment over time, and the volume of news. Then he presented an outline of how the concepts of log optimal growth and ‘volatility pumping’ can be introduced in news enhanced trading strategies.
An Automated Trading Strategy for Equities Using Crowd-Sourced Earning Data was described by Ilya Gorelik from Deltix. He explained well the implementation of an automated equity trading strategy based on aggregated company earnings estimates sourced from independent, buy-side and sell-side analysts, along with those of private investors.
Day two of the conference was opened by Stephen Pulman, Professor of Computational Linguistics, Oxford University / TheySay (Analytics). The keynote speaker highlighted “Compositional Sentiment Analysis”.
Louis Scott, Northfield Info Services Inc and Founder of Kiema Advisors presented “Sentiment, News and Volatility in both Stock Returns and Trading Volume”. He explained well, how Northfield uses alternative paradigm to examine the role of sentiment analysis.
Social Media Analytics Applied to Finance was explained well by Sri Priya Ponnapalli from Bloomberg. She discussed as social media is moving into the financial sphere, Bloomberg can evaluate the opinions about traded instruments and track news and sentiment with respect to e.g. equity issuing companies in real-time.
Another interesting aspect of the conference were the panel discussions spread over three panels: Panel 1:- Adoption of Sentiment Analysis in Fund management & Trading Strategies, moderated by Gautam Mitra from OptiRisk Systems; Panel 2:- Use of News and Social Media Data for Market Surveillance & Operational Risk Control, moderated by Ashok Banerjee from IIMCal; Panel 3:- Data Sources: Market Data, News (Meta) Data, Social Media Data, Google Trends & Others, moderated by Andrew Rummer from Bloomberg. The panels engaged the participants in engaging interactive discussions.
Bing Liu, a Professor and subject expert from University of Illinois at Chicago delivered the closing keynote talk: “Sentiment Analysis: Past, Present and the Future”.
OptiRisk a company specialising in financial analytics is compiling the update of their Handbook of News Analytics in Finance. Contributions are solicited for the New Handbook: Sentiment Analysis Applied to Finance. Please see the link: www.optirisk-systems.com/publications.asp for more information.
Also the conference recordings and the slides can be obtained by making email requests to: firstname.lastname@example.org.
We thank all the active participants of the conference in particular a big thank you to our sponsors Bloomberg, Thomson Reuters, RavenPack, OptiRisk Systems, Northfield Information Services, Deltix.
UNICOM is planning the repeat event “Behavioural Models & Sentiment Analysis Applied to Finance” in summer of 2015 in London. ..WATCH THIS SPACE.