Building a Retail Equity Trading Model Using Social Media Posts w/ Art Steinmetz- nyhackr Nov Meetup

Visit https://www.nyhackr.org to learn more and follow https://x.com/nyhackr

Talk Title- Building a Retail Equity Trading Model Using Social Media Posts

Talk Description- We examine a publicly available dataset of over 100 million social media postings about stocks between 2010 and 2022 to identify users who made statistically valid predictions about future stock performance.  We find that a tiny fraction of users make statistically valid predictions, though they are often wrong.  There is as much value in predictably bad performance as good performance.  We go on to build a trading strategy that a small retail stock trader could implement based on these findings.

From a data science perspective, this is interesting because we use the R package, "Duckplyr" and DuckDB to quickly and efficiently manipulate a multi-gigabyte data set either on a local disk or from an Amazon AWS S3 bucket.  The project was inspired by this paper, "StockTwits: Comprehensive records of a financial social media platform from 2008 to 2022" by Li, X., Al Ansari, N., & Kaufman, A. (2025)

Bio- Currently, Art Steinmetz is a private investor located in New York City and an angel investor in several startups including:
-Tifin AI for wealth management.
-Market Reader Understand markets in real time.
-Stronghold Payment and financial infrastructure for all.
-Launchpoint Advanced air mobility.

Additionally, he serves on the board of the Blackrock fixed income funds and do consulting for an enterprise data science firm, Posit PBC. Art occasionally publishes recreational data science projects at outsiderdata.blog. He is a member of the Economic Club of New York.

Art only considers investment introductions from my established network. Unsolicited inquiries will be politely ignored.