My Kingdom for a Tweet…

by Joshua Tucker on April 24, 2013 · 6 comments

in Social Media


This is the fake tweet (the AP twitter account was hacked) that yesterday caused the US Stock market to briefly lose $200 billion worth of value. While the market recovered afterwards, I take this as yet another reminder that social media are not merely mimicking functions played by traditional forms of media. This incident points to two crucial aspects of divergence. First, the speed at which information can spread across social media networks is truly stunning. Note the time on the tweet: 1:07 PM. Now note the time of the market drop:


Second, this event took place because someone hacked the AP twitter feed (according to the Wall Street journal “a group of enthusiastic Syrian youths” have claimed responsibility). It’s difficult to imagine anyone hacking the front page of the NY Times or brainwashing Walter Cronkite. Moreover, while hacking the AP account certainly enhanced the number of followers who saw the tweet and enhanced it’s credibility, it is not difficult to imagine this sort of “hoax” spreading quickly from a home grown Twitter account either, especially given the fact that plenty of people access Tweets via #hashtags (keywords) in addition to simply following others. At its core, social media continue to be differentiated from traditional media by the fact that anyone can send a message. How many people will listed to that message will of course vary, but send the right message in the right circumstances that looks the right way, and you can knock $200 billion out of the US economy in a heartbeat.

[h/t to Per Dutton.]


Daniel April 24, 2013 at 8:55 am

And then the stock market immediately bounced back to right where it was before. Why does this matter?

Jose C Silva April 24, 2013 at 10:13 am

Possibly much of the reaction speed here is a product of automated trading based on text analytics tracking social media (especially since AP is a reputable news source).

Scott Monje April 24, 2013 at 10:19 am

Also interesting are the commentators who, seeing the stock market reaction, immediately “knew” that this had been done expressly for purposes of short selling.

Matt Belcher April 24, 2013 at 10:38 am

I work in a futures trading shop so I’ve got access to the real-time market data for the e-mini S&P futures on the CME. The tweet appeared at 13:07, but the market didn’t move until 13:09:30 or so. The reason the market moved then is the tweet was picked up and repeated by RANsquawk ( at exactly that time. I’m sure there are some traders out there looking at the AP’s twitter feed, but not nearly as many as listen to RANsquawk.

Joshua Tucker April 24, 2013 at 5:45 pm

Matt: Thanks for this comment – this is really interesting! I was in a separate discussion with a friend about the role of computers vs. humans in processing this information. Your answer hear suggests that the key is humans (although possibly the Ransquawk feed is automated?). How long until you think humans are out of the link altogether, with machines directly mining social media and converting that to buy and sell orders? Or is RANsquawk already run by an algorithm? Thanks!

Matt Belcher April 26, 2013 at 1:55 pm

At least in this case the humans are the key. RANsquawk isn’t an automated feed.

I’m sure there are some algorithms which use twitter feeds as inputs, but they aren’t big players yet. Most of the news-driven algorithms run on timed news releases. For example, you can buy a news feed from Bloomberg that will give you real-time versions of the BLS releases, etc. These feeds cost > $10k a month last time I checked but they are worth it if you can be the first person to react to the unemployment number. Unscheduled news events are a lot harder to automate because they could be anything and it isn’t clear which way the market should move in response. You’d also have to make sure that whatever source your algorithm uses as input really is first by measuring recent market moves to make sure you aren’t just reacting late.

Comments on this entry are closed.

Previous post:

Next post: