Can ChatGPT decipher Fedspeak?
That was the question a study explored recently. The study wanted to find out how much financial language the AI could understand when compared to humans.
In particular, they focused on ‘Fedspeak’, the jargon used by the Federal Reserve to announce monetary policy decisions.
If you’ve heard Fedspeak, you know it’s these technical, vague, and long sentences. In the words of former Fed Chair Alan Greenspan:
‘Opaque answers to straightforward questions were part of the job, because I couldn’t say “no comment” and didn’t want markets to overreact.
‘What tends to happen is your syntax collapses. All of a sudden, you are mumbling. It often works. I created a new language which we now call Fedspeak. Unless you are expert at it, you can’t tell that I didn’t say anything.’
Granted, the Fed’s communications have gotten better over the last few years…but plenty of people still pore over every Fed statement, checking for slight changes in wording to understand what the central bank is thinking.
But back to the study.
The researchers chose hundreds of sentences from Fed policy statements that were published between 2010 and 2020, and then asked the AI to sort those statements in different categories that ranged from dovish (policies that support low interest rates) to hawkish (support for higher interest rates).
So, what was the result?
While the study showed that the AI can still make mistakes and fail where a human may have done better, the researchers found that:
‘GPT models demonstrate a strong performance in classifying Fedspeak sentences, especially when fine-tuned.’
They noted that the AI was able to provide reasons for its choices and what’s more, as the researchers said, it was ‘very similar to the reasoning provided by a human reviewer’.
I mean, that’s pretty impressive.
AI’s edge is speed
This follows another research study that used sentiment analysis to explore whether ChatGPT could predict how a stock price would move based on a news headline. Long story short, they found that ChatGPT ‘outperforms traditional sentiment analysis methods’.
As the researchers concluded, the chatbot would be great as a complement tool for trading strategies:
‘Our results suggest that incorporating advanced language models into the investment decision-making process can yield more accurate predictions and enhance the performance of quantitative trading strategies.’
That is, it looks like these technologies could look at statements and news quicker than a human to help make quick trading decisions.
While there’s plenty of hype — and fear — about what AI can do, it’s true that these tools are getting better.
And AI’s big advantage is speed. These systems can process massive amounts of data, statements, reports, and news stories in an instant and learn on the go.
That’s why, when it comes to finance, some of the biggest names in the industry have been investing in AI and data technologies for years to be able to compete.
As more people have access to AI tools, this will not only disrupt finance…but every sector of the economy…and increase productivity.
Goldman expects that as more people in the economy use natural language-processing tools, like ChatGPT, we could see a 7% increase in global GDP and a 1.5% increase in productivity over the next decade.
So, for all the fearmongering out there on a coming recession, this is pretty positive…
AI is part of our investing toolkit
Even here, at Fat Tail, we’re using AI for investment.
My colleague and friend, Callum Newman, is a big believer in AI. And he has been using an AI system as part of his stock-picking method for quite some time in the more speculative side of the market, small caps.
As Cal recently put it, it’s made his stock-picking process much quicker:
‘What our system does is incredible. It uses artificial intelligence to filter stocks down from 2300 equities on the Aussie market to a handful of high potential trading opportunities. This process helps me assess the risk in each trade I recommend much more quickly, which is a major advantage when you are finding small cap stocks to trade.’
In fact, he’s holding a free online event tomorrow, where he will show you exactly how it all works. What’s more, during the event, Cal will even give out a free trade recommendation his AI system has flagged, including the name, stock ticker symbol, and buy-up-to price. All you need to do is tune in to the presentation to get the details.
So, if you’re interested in short-term trading, make sure to check out Cal’s event, ‘The MAT Advantage’, tomorrow.
Best,
 |
Selva Freigedo,
For Money Morning
