Algo trader and investor sensitivity to the trade war headlines should have increased before the Trump-Xi meeting in Japan. This is because, not only the number of mentions has risen, but their importance has increased too, helping to get additional precision in calibrating expectations of the possible trade talks outcome.

Here is a spectacular example of stocks and the Yuan’s “clueless” mini-pump, based on a “bullish headline” from CNBC, which quoted the head of Treasury Department Stephen Mnuchin, adding a minor grammatical mistake which turned into a gross semantic error.

According to the CNBC headline:

“U.S. Treasury Secretary Steve Mnuchin says U.S.- China Trade Deal is 90% Complete”

Perhaps many of you saw this headline yesterday, and my first reaction was bewilderment, as it was totally at odds with tariff threats, company blacklists and many other signs of continuing tension. I think many will agree that moving forward with the agreement requires at least a truce. The headline would make sense considering past information, like when the White House announced that the deal was almost ready, but China sent the draft with significant amendments. Bringing talks back to square one and Trump firing off new tariffs, etc. It was not clear why Mnuchin was talking about it in the present tense. Nevertheless, the headline was enough for algos to start pumping:

1-15.png

2-15.png

In addition to the article on the website, CNBC also posted the news on Twitter, which is even more popular with algorithms.

Later it turned out that CNBC made a minor grammatical, but a gross semantic error! With Mnuchin’s actual reply being the following:

"We were about 90% of the way on a China trade deal and there’s a path to complete this".

Mnuchin, as usual, did not say anything new and chose to reformulate his own words and other common knowledge. On the charts, you can see how the Dow futures and the offshore yuan later erased gains from the pump.

Unfortunately (or fortunately), we can say that algo software employs fairly naive language processing approaches and can be easily confused, causing “dumb” market manipulations.