61.3 F
Los Angeles
Tuesday, Dec 5, 2023

Media Bias

A Century City startup claims to have the answer for traders frustrated by the notoriously unpredictable stock markets.

Quant the News LLC, founded by local entrepreneur Brett Markinson and techie Ben Goertzel, recently unveiled a Web site aiming to give quantitative analyses of news articles that investors can use to make investment decisions.

Known as Stockmood, the site uses sophisticated formulas to analyze aggregated online news feeds to determine whether articles written about a particular company are generally positive, negative or neutral. It then compares that coverage to the company’s share price and assesses the “mood” of a stock, which can be used when making decisions to buy or sell.

If a stock is in a bad mood, negative news has pushed its price down, while a good mood means the opposite. However, just because a stock is in a good mood doesn’t mean the Web site will recommend that you should buy shares. The algorithm may determine that the stock may be overhyped or near a peak and ready for a drop.

“We look at this as a new kind of financial indicator,” said Goertzel, who is based in Washington, where he has developed artificial intelligence systems through his technology firm Novamente LLC. “We think (investors) should be looking at the quantitative news sentiment and the quantitative market mood.”

While some hedge funds and institutional investors have toyed with similar but costlier services, the founders say Stockmood is the first site to offer such a service free of charge to the average investor.

Stockmood is currently up and running in beta form, and the pair plan a larger rollout in early 2009. The duo says they already have “a couple thousand” registered users.

Though they will continue to offer a free service through the site, Goertzel and Markinson said they plan to make money by offering a more full-featured version for a subscription fee.

But before investors are willing to plunk down money, most have one question: Does it work?

According to the pair, it does. Sort of.

“It depends on what ‘work’ means,” Goertzel said. “Are we accurately telling the sentiment of the news? Our AI is correct around 80 percent of the time. What’s the predictive value in terms of making money? That comes out to be a subtler question because there are lots of different ways to use this sentiment.”

Indeed, the service is not intended to simply tell investors to buy Stock A and sell Stock B. Instead, Goertzel said, Stockmood is a guide to help investors make general decisions about potential investments, but there is always uncertainty and risk.

Goertzel and Markinson have been testing the system for more than a year to see if it could have helped them come out on top.

Though he would not provide details, Goertzel claimed, “It would definitely have made money.”

That would be appealing for investors concerned about the recent turmoil in the markets. The Dow Jones industrial average recently closed below 8,000 for the first time in more than five years as the stock markets continue a prolonged plunge.

But is this a good time to launch a stock services Web site, when the recent bear market has even investors with the soundest stomachs bailing out? Perhaps.

E*Trade Financial Corp., a New York-based brokerage, said it had a net gain of 63,000 accounts in October despite a $22.8 billion drop in customer assets from September. That is good news for services such as Stockmood that cater to individual investors.

“Our customers are still highly engaged,” said Tina Martineau, a spokeswoman for E*Trade, adding that the company does not offer any services like Stockmood.

Starting out

Many investors look to the financial statements and guidance provided by a company when deciding to buy a stock. But Markinson, who describes himself as “a passionate investor in the markets,” realized that stocks are often impacted very directly and measurably by news reports.

“I’d had this idea about integrating news into the trading system,” he said.

Markinson had only toyed with the notion until he met Goertzel through a mutual acquaintance. Armed with a PhD in mathematics, Goertzel has developed artificial intelligence-based trading systems for hedge funds and has created software for the government.

With Goertzel’s technological savvy and Markinson’s trading background, the pair had little trouble finding investors. After securing $150,000 in seed financing from Markinson’s venture capital firm Lagovent Private Equity Group LLC and Matt Coffin, the founder of Santa Monica-based LowerMyBills.com, they started developing the program.

They enlisted several people to read thousands of news stories and determine whether the articles conveyed a positive, negative or neutral sentiment. Meanwhile, they developed a program that could mimic the judgments of humans. It essentially uses advanced formulas to identify words or phrases that indicate a positive or negative sentiment in financial news stories, said Goertzel.

Though the system has proved fairly accurate, the pair nevertheless decided to give it a feature that allows it to correct errors. If a user believes Stockmood’s assessment of an article’s sentiment is wrong, that person can click a button to register the error. If a number of people believe a classification is wrong, the system will correct itself. What’s more, the system “learns” from the correction and will incorporate the knowledge into future assessments.

To accurately determine the “mood” of a stock, the site must have a large number of articles to analyze. Therefore, Stockmood only works now for 135 larger companies, such as Walt Disney Co.

On a recent day in November, Stockmood issued a “high pessimism alert” for Burbank-based Disney, meaning shares had fallen on bad news and were likely ready to rise. The next day, shares jumped more than 7 percent.

But some experts caution against reading too much into news sentiment when making investment decisions. Edward Wedbush, president of Los Angeles-based investment bank Wedbush Morgan Securities Inc., said there are inherent risks in using the media to guide investment decisions.

He pointed to a recent blunder that hurt shares of United Airlines’ parent, UAL Corp., as a reason to be cautious. On Sept. 8, Bloomberg News mistakenly distributed a six-year-old article about the airline’s 2002 bankruptcy filing as breaking news. Panicked investors sold off the stock in droves, driving shares down 75 percent before the error was revealed.

“In the news media, you always have the (risk) of lack of accuracy,” Wedbush said.

But he said this kind of service could be beneficial if it can avoid such mistakes. “It’s probably OK to be using that kind of information as long as there are safeguards.”

Quant the News LLC

Headquarters: Century City

Founded: 2007

Core Business: Providing investment advice using artificial intelligence technology that analyzes news stories

Employees in 2008: 2

Goal: To become a widely used tool for casual and professional stock investors

Driving Force: The desire of stock investors to find a trading edge

Featured Articles

Related Articles