Technology

AI/share trading: earnings call bingo


Players of Buzzword Bingo check off jargon spouted by chief executives in hopes of scoring a full house. It has long been a pastime of bored employees at mandatory “town hall” meetings. It is increasingly played by robots whose human handlers aim to make a killing on the stock market.

In theory, a computer should be able to pick up clues to profits performance by screening screeds of verbiage. Earnings calls are among the biggest potential data mines. Here, bots dig for price clues amid such ballast as analysts uttering the time-honoured ice breaker “great quarter, guys!”.

For the theory to work, conditions apply. The AI program needs some linguistic ability. Words such as “competitive” have good or bad meanings dependent on context. The utterances of chief executives only matter if they have a clue where their business is headed. Many do not.

Even so, hedge funds are increasingly running algorithms on verbal data sets. The uniformity of boss-speak suits machine learning techniques, they say.

Lex made an empirical analysis of two sets of calls. The first set involved two UK companies. These were retailer Superdry and insurer Saga. The ratio of upbeat to downbeat words fell sharply before profit warnings at both groups.

We also screened Coca-Cola’s first-quarter calls. Words such as “price”, “mix” and “volume” were used frequently last year. This year, “currency” and “markets” abounded. A jump in the use of “innovation” stood out. Might the business fear its sugary drinks are going out of fashion?

More scientifically, Goldman Sachs screened 4,000 earnings calls from S&P 500 companies over the past year. The results paint a starkly different picture to the one portrayed by stock prices near record highs. Goldman’s index of earnings call sentiment was at its lowest level in June. Concerns over slowing growth were high. Goldman found a “significant relationship” between the mood of earnings calls and subsequent stock price movements.

Luddites will wonder why investors cannot simply listen to earnings calls themselves to determine how chief executives feel. But these Q&As can last over an hour. And they can be very boring.

Finally, as a point of fact, analysts rarely say “great quarter, guys!” S&P data show the phrase was only uttered twice during six months of US calls. Well done, Matt from JPMorgan and Bill at DA Davidson. You upheld a dying tradition.

The Lex team is interested in hearing more from readers. Please tell us what you think of language-based quant analysis in the comments section below.



READ NEWS SOURCE

This website uses cookies. By continuing to use this site, you accept our use of cookies.