White Papers Making more of data using AI | Page 6

Data Management

A ComTech Advisory Whitepaper

A number of software products already exist to help traders make use of social media 3, including
Reuters and Bloomberg. These tools and services are mainly targeted at equities traders and even
include tools to automate trading based on sentiment analysis derived from social media and the
blogosphere. University research in 2010 showed that social media mentions of a brand tended
to correlate with its share price. Other types of social media can also be used in this way including
the number of ‘likes’ on YouTube and Facebook, for example. However, can commodity traders
utilize these same social media resources as well?
Social media, like Twitter, along with conventional news sources and unconventional news
sources like blogs, can be used to provide sentiment analysis and buy/sell signals that might prove
useful especially when used in combination with traditional data. Part of the issue is the vast
amount of ‘noise’ that needs to be filtered out as well as an ability to know what keywords to
track and where to track them. Without appropriate tools, this task would be somewhat
impossible or at least not very cost effective.

Artificial Intelligence
One way to solve this problem is via Artificial Intelligence (AI) using Natural Language Processing
(NLP) and Complex Event Processing (CEP). Both NLP and CEP have not been used extensively to
date as a data management tool in the commodity trading arena and DataGenics’ survey showed
that just under a third of those interviewed had never used AI tools but almost half could see it
having value. However, as the volumes of data increases having computers detect patterns and
categorize data according to rules
Figure 4: The Value of AI
set by the operators is one way to
High Value
make better use of the data and
6%
highlight information that might
ordinarily have been missed. If you
25%
Moderate
involved
the
vast
volumes
Value
associated with social media of any
0%
38%
Low Value
kind, then the need to utilize
Artificial Intelligence is essential. A.I.
is already used on Wall St. to
31%
None
develop trading sig