A founder built TickerTags by hand-curating 1.5 million word combinations from Twitter to track how people discuss publicly traded companies. The system flags unusual spikes in mentions that could move stock prices, benchmarking real-time data against historical patterns and seasonal norms.
Me and my business partner created a platform called TickerTags in the mid-2000s with Twitter, and we had access to the Twitter decahose, which is a 10% randomized sample of every tweet in real time. And we hand-curated about 1.5 million word combinations that represented how people were speaking about every product, brand, basically anything that was connected to any publicly traded company or meaningful to any publicly traded company in any way. We had organized into a taxonomy. So every company had like 300 to 1,000 combinations of words. So we were actually monitoring in real time the frequency of mentions of those 1.5 million words, benchmarking them against historical norms, including seasonality. And so since it was organized in a taxonomy, when there was any type of anomaly in speech patterns happening across Twitter that were impacting a subject matter that we had curated to be potentially impactful to a publicly traded company, System would flag that. And so that was basically me taking my methodology of what I had done manually and institutionalizing it for Wall Street.
