Finance professor at University of Toronto. Interested in securities exchanges, technology, and investments.
Prepared by: Marius Zoican
3rd Toronto FinTech Conference November 5, 2020
"The Social Internetwork and Stock Returns"
(Al Guindy and Riordan)
Identify pervasive "themes" in the economy
How do you think of Tesla?
- A car manufacturer (like Ford) or
- A tech company (like Facebook)
- Use Twitter cashtag data (e.g., $TSLA) to identify how people make connections between firms.
- Very refreshing and innovative idea!
Thematic investment is here already
- ETFs invest in cross-industry themes (e.g., religion)
- "Top-down" themes: proposed by ETF issuers.
- This paper: "bottom-up" -- what do investors perceive?
Social Internetwork is to Thematic ETFs as
Principal Component Analysis is to Factor Identification
In essence, an agnostic & data-driven perspective of what the themes might be:
(useful for ETF design?)
Next step: assess marginal contributions?
Interesting to decompose the social inter-network.
How much of it (on the margin) comes from:
- Industry peer correlations;
- Supply chain relationships;
- Corporate governance / ESG values;
- "Intangible" perceptions: Tesla is a tech company.
- Technical analysis and "tips" from Twitter users.
You mention many times "non-linear, non-trivial" connections. Next step is to flesh them out!
How much of the data is noise?
is a Wild West!
I would narrow it down. Can you distinguish between:
- Corporate Twitter accounts;
- Financial industry Twitter accounts;
- Journalist Twitter accounts;
- Everyone else.
- What does the strength of a link represent? Re-tweets? Commonality of perception?
- Why not control for traditional factors in the return analysis (Fama French, momentum, etc.)?
- Are some companies much more likely to draw attention from social media users (cool, flashy Tech stocks)?
- To what extent does the network capture market sentiment?
Discussion: The Social Internetwork and Stock Returns
By Marius Zoican