NPR logo

Trading Bot BOTUS Will Buy And Sell Stock Based On Trump's Tweets

  • Download
  • <iframe src="https://www.npr.org/player/embed/523890750/523890751" width="100%" height="290" frameborder="0" scrolling="no" title="NPR embedded audio player">
  • Transcript
Trading Bot BOTUS Will Buy And Sell Stock Based On Trump's Tweets

Trading Bot BOTUS Will Buy And Sell Stock Based On Trump's Tweets

Trading Bot BOTUS Will Buy And Sell Stock Based On Trump's Tweets

  • Download
  • <iframe src="https://www.npr.org/player/embed/523890750/523890751" width="100%" height="290" frameborder="0" scrolling="no" title="NPR embedded audio player">
  • Transcript

Planet Money built a stock trading Twitter bot. The big puzzle is figuring out which company to buy or sell, and how to get a computer to tell the difference between Apple the company and the fruit.

DAVID GREENE, HOST:

So all this week we have been looking into a trend on Wall Street. Stock market trading is becoming more and more automated. To understand this better, our Planet Money podcast team is building what is called a trading bot, their very own automated stock trading computer program. Here's Alex Goldmark.

ALEX GOLDMARK, BYLINE: We've got the plan down. Our bot will read the tweets of President Donald Trump and then use that to trade stocks. When he says something positive about a company, we buy that company's stock. When he says something negative about a company, we will sell that company's stock short. We will set ourselves up to make money if the stock price goes down.

We teamed up with a hedge fund called Tradeworx to actually build the bot. One of the computer programmers there, Camilo Jimenez, he got to work. He did what he usually does when he's making a bot - researching from his home office on a little wood desk in his bedroom, keeps it simple.

CAMILO JIMENEZ: Just my laptop, coffee and notepad. That's basically it.

GOLDMARK: Laptop, notepad and some coffee, that's what you need.

JIMENEZ: Yeah. That's all you need.

GOLDMARK: That's where he hit a problem. His project was to get the computer to recognize which company Donald Trump was talking about.

JIMENEZ: That's very hard. That's easy for a human. That's very hard for a computer because the text could include either the name of a person or the company or the product and you have to link that to a company.

GOLDMARK: So is Donald Trump just saying he ate an apple for lunch, or is he talking about Apple computers?

JIMENEZ: Correct. That's a really good case.

GOLDMARK: Camilo had to train our bot to tell the apples apart, so he added in what's called a context algorithm. He had our little bot read a whole bunch of financial articles about Apple the company and learn the kinds of words and phrases and topics associated with the company - Steve Jobs, iPhone, fanboy - and then learn the things that are not associated with the company but with apple the food - eating, juice, Honeycrisp.

And the machine, it really is learning about apples - both kinds - and then about every company. But there was one company that the bot could not recognize. The company name is a name that President Trump tweets about. It's something pretty close to him but the bot just kept getting confused. You want to guess what it was? I'll give you a sec - Tiffany's, the jewelers.

MANI MAHJOURI: In our production algorithm, we won't trade the company Tiffany's because the computer can't tell when Trump is talking about his daughter and when he's talking about the actual company.

GOLDMARK: Mani Mahjouri is the head of investments at Tradeworx. And he also helped us build our bot. He said with enough work, they could train the bot to recognize Tiffany's the company from Tiffany's the daughter but it's just not worth it. He made a call. Our bot, it doesn't trade Tiffany's.

And that's just one of many human judgments that's underneath our automated stock-picking robot. Trading with bots is a blend of humans and computers making decisions. For us, it adds up to a bot that can pick out positive and negative tweets. And it can figure out what companies the president is tweeting about - except for Tiffany's. So it's ready to go, pretty much.

MAHJOURI: You know, we're - we feel good about the strategy, you know. And...

GOLDMARK: Can you say that with a little more confidence? You're...

MAHJOURI: Can't.

GOLDMARK: ...Saying that to me like...

MAHJOURI: I can't. No, because...

GOLDMARK: ...You really don't believe it.

MAHJOURI: No, no, no. I just don't want to understate the risk. And I don't want to overstate the benefit of doing something like this. I just feel like, you know, it's not - we haven't scientifically proven anything.

GOLDMARK: But he said if you wait for proof, you can miss out on a good thing. So we're doing this. And we're going to do it with real money. The staff of the Planet Money podcast is putting our own personal funds behind the plan, not NPR's money, our money. And we gave our bot a name - BOTUS, Bot of the United States. Now, we wait for the tweets and the trades. Alex Goldmark, NPR News.

(SOUNDBITE OF ARMS AND SLEEPERS' "CAN YOU - INSTRUMENTAL")

GREENE: BOTUS like POTUS, I see what you guys did there. BOTUS is going to be tweeting about any trades it makes. You can follow BOTUS on Twitter @botus or at npr.org/botus. That is B-O-T-U-S.

Copyright © 2017 NPR. All rights reserved. Visit our website terms of use and permissions pages at www.npr.org for further information.

NPR transcripts are created on a rush deadline by Verb8tm, Inc., an NPR contractor, and produced using a proprietary transcription process developed with NPR. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.