While the stock market may seem anything but predictable, a study by University of Connecticut researchers that mined and analyzed Twitter data has helped them develop a stock option forecasting model that tripled investors’ profits when applied to simulation portfolios.

The study, “Twitter volume spikes and stock options pricing,” published recently in the journal Computer Communications, reveals how spikes in the number of tweets about a company can be used to design a profitable stock options trading strategy.

“Our study is the first that analyzes the relationship between Twitter volume spikes and stock options pricing,” says Bing Wang, associate professor of computer science and engineering, one of three authors on the paper. Co-authors were Wei Wei of FinStats.com, a big data analysis company, and Yuexin Mao, a UConn Ph.D. student in computer science and engineering advised by Wang.

“Our results show that social media is a powerful tool to help understand the behavior of stock options, and further assist the trading of these valuable, but complex investment vehicles,” says Wang.

… a collection of tweets reflects the collective wisdom of the users who post them …
— Bing Wang

Since Twitter’s creation in 2006, the San Francisco-based short messaging service has grown to more than 500 million users, more than half of them active users. Traders, investors, financial analysts, and news agencies, for example, post tens of thousands of tweets daily related to stocks, making Twitter the largest knowledge base of free investment advice about the financial markets that is available to everyone instantly and easily.

“Since a collection of tweets reflects the collective wisdom of the users who post them, a Twitter volume spike about a company may contain important information about the stock,” notes Wang. “Our goal was to investigate whether Twitter volume spikes could shed light on the behavior of stock options pricing.”

Options are created when a company launches new shares of stock; options market traders then begin pricing and trading contracts to buy or sell the new stock at various prices in the future. As more investors have learned about the versatility of options – big profits can be made when the stock goes up, but also when it goes down – options trading has become an increasingly popular way for active traders to make bets on the direction a stock may take, or to hedge existing positions in their portfolios.

This trend has been made possible by the advent of electronic trading and data dissemination. Computers have enabled unprecedented insights into complex securities by empowering traders, using algorithms, to sort through reams of commercial data – such as Twitter – to accurately measure, evaluate, and predict changes in the price of financial assets.

An option gives investors the right, but not the obligation, to buy or sell a stock at a specific price on or before a certain date. There are a wide variety of option contracts, both short-term and long-term, but all essentially have two sides:

• A call option grants the right to buy the stock at the specific price, called the strike, up to the expiration date. Buyers of calls hope that a company’s earnings will drive up its stock price before the expiration date.
• A put option confers the right to sell the stock at the strike price. Put buyers hope the stock price will fall before the option expires. For example, if you own a Standard & Poor’s 500 Index fund, you can buy put options that increase in value if the market declines.

Regardless of their call or put objective, options are risky, as investors must juggle large numbers of variables when calculating the potential of risk or reward. These variables include the magnitude of the stock’s movement; the direction of the movement; and the timing of the movement, among others.

Today there are sophisticated trading tools to help track variables that drive option trades; still, the complexity of this data stacks the odds against traders. Using mathematical models that estimate the price of call and put options over time, the UConn researchers monitored and analyzed Twitter data to identify option trading opportunities.

To obtain data for the study, the UConn team used the closing price for stocks in the S&P 500 index spanning August 2013 to August 2014. Using algorithms to sort through the deluge of daily tweets, 3,288 Twitter volume spike samples were identified – at least one for each company represented in the S&P 500 index – which were then used in the analysis of stock options pricing.

The study determined that extreme stock price swings were correlated with Twitter volume spikes, as sharp increases or decreases in returns triggered more discussions about them, and hence more tweets. The researchers reported that stock prices are very volatile in the days around the stock’s Twitter volume spike, particularly the days immediately before and after the spike.

They also found that stock options may still be overpriced right after a Twitter volume spike; and that put options tend to be priced higher than call options. Wang says the results indicated that Twitter volume spikes could be used as a trading signal to sell overpriced put options right after the spike in order to gain profit.

Based on their findings, the researchers created a simulated portfolio using one year of stock market data to evaluate their put selling strategy. They found that even in a conservative setting, this strategy achieved a 34.3 percent gain, while the S&P 500 Index only increased 12.8 percent during the same period.

Recognizing the potential that Twitter data offers for predicting stock volatility, Wang says she believes a more detailed analysis of Twitter data, such as analyzing the content of tweets, can provide even more insights.