Prattle is a Fintech innovator whose system started by forecasting market reactions to central bank communications. As a member of the BasisTech startup program Prattle uses Rosette entity extraction to detect the speakers in central bank communications. They’ve since expanded to power analyses like the blog post below:

Despite what the financial press might want you to believe, Trump’s tweets about specific companies only result in 0.3% Cumulative Above Average Return (CAAR). However, when the President tweets about a specific person at a company (such as Rex Tillerson of Exxon) the CAAR jumps to 9.2%. It also appears that these corporate officer mentions have spillover effects across the entire industrial sector.

In addition to examining company/officer mentions in tweets, Prattle also ran a comparison of the impact of company/officer mentions in Trump speeches vs his tweets–and tweets were the big winner. Trump speeches caused a 2% CAAR; his tweets generated a 3% CAAR.

Prattle quantifies the market impact of language with Rosette text analytics under the hood, performing named entity recognition and linking entities to known Wikidata. While we’ve traditionally used Rosette to help us understand the impact of corporate and central bank communications, we’ve also found it adept at understanding the market impact of tweets. For each one of Trump’s tweets from the last year that mentioned a corporation or a corporate executive, we performed a market-model event study centered around that tweet. Each study precisely linked Trump’s words to market movement. We computed the CAAR by combining the results of these studies.**

Our research suggests that media attention paid to Trump’s twitter account is well deserved…but imprecise. The real market effects of those tweets occur when he names specific corporate officers, not when he mentions the corporations themselves. Investors would be wise to discern signal from noise and focus on tweets that mention corporate officers.

For more content that breaks down the market impact of the world’s most important communications, subscribe to Prattle’s blog, The Signal. If you’re interested in adding Rosette to your analytics stack, try out Rosette API for free, or talk to the sales team.

* This is a composite figure for tweets about both companies and corporate officers.

** We excluded mentions of news entities. Event studies benchmarked on the S&P, with a 40-day bandwidth centered on the tweet and a 170-day baseline regression.