Using Social Media to Predict Stock Performance

What if you could use social media to make accurate predictions about the success or failure of a company?

Better yet, what if you could use social media to predict the performance of individual stocks?

As it turns out, you actually can. A growing number of studies are finding evidence that using social media can be one of the most surefire ways of pinning down companies that outperform their peers.

Now, just to be clear, I’m not talking about Facebook or Snapchat here. Sharing baby pictures or posting memes obviously isn’t going to give you the upper hand when it comes to picking winning stocks.

However, there are various ways to use sites like Twitter, Yelp, and Glassdoor as supplemental tools for market research.

Most of the information coming from these social media platforms is just noise, but if you know where to look, they just might provide you an unfair advantage.

Let me show you what I mean…

Legal Insider Trading: Glassdoor.com

Glassdoor is an online review site where current and former employees can anonymously review companies and management. Glassdoor was launched in 2008 and has already brought in over half a million reviews for more than 371,000 companies across the U.S.

Because of this, Glassdoor arguably is the single-best indicator of a company’s culture and morale. What better way to understand how things are going at a firm than by listening to those who used to or currently work there?

Glassdoor allows its users to rate companies on a variety of different metrics. This includes culture and values, compensation and benefits, career opportunities, work-life balance, hiring practices, and CEO approval, to name just a few.

Among these metrics, there are a select few that have proven to be particularly telling indicators in terms of stock performance. We’ll look at one example:

Considering a CEO is the head of a company, you might expect CEO performance to have a strong correlation with the company under his or her wing. As it turns out, this holds true.

Glassdoor conveniently lists its top 50 ranked CEOs according to user ratings. We took all publicly traded companies within the top 20 and looked at stock performance over the last five years.

We then compared the returns to the S&P 500 and Dow, finding that these companies do indeed tend to outperform the broader market.

Over the last five years, the S&P and Dow have returned 58.9% and 45.4%, respectively. On average, public companies within Glassdoor’s top 20 ranked CEOs returned 82.7%.

Further, two-thirds of the companies we looked at outperformed the S&P and Dow. Only two ended up with negative returns:

Click Image to Enlarge

Consumer Knows Best: Yelp

Most people are already familiar with online review site Yelp. It’s a great tool for figuring out where to find the best nearby sushi joint. It may be an even better tool for figuring out which companies to add to your portfolio.

In December 2014, researchers at the University of Maryland analyzed Yelp reviews for 2,000 open restaurants and 450 closed restaurants in the Washington, D.C. area. In total, the researchers mined data from over 130,000 reviews and a total of 15 million words.

Specifically, these researchers looked for specific linguistic patterns and words. What they found is that certain words and phrases act as strong predictors for which restaurants stay open and which go out of business.

For instance, restaurants containing reviews consisting of the words “food,” “good,” “place,” “like,” “order,” “friend,” “time,” “great,” “nice,” and “service” tend to have unusually high survival rates. According to researchers, the correlation for many of these words and phrases is significant enough for them to act as predictors to business success.

Now, the reasons for this should be somewhat obvious…

First, good reviews reflect good business practices and high product quality. Likewise, bad reviews reflect poor business practices and low product quality.

Second, online consumer reviews end up driving demand. The higher a business is rated, the more customers it attracts. The lower it’s rated, the more customers stay away.

According to a recent publication from Harvard Business School titled, “Reviews, Reputation, and Revenue: The Case of Yelp.com,” for instance, a one-star increase in Yelp rating leads to between a 5% and 9% increase in revenue. That’s a pretty big deal.

Of course, it’s important to keep in mind these studies looked only at restaurants, but we think it’s safe to assume the same effect would hold true across all kinds of businesses.

Likewise, using consumer reviews to predict stock performance extends far beyond Yelp. Amazon product reviews, for instance, could be used as indicators for how well a product will perform (there’s a reason the iPhone 6 has a near five-star rating).

Social Media Analytics

Finally, there’s an entire field of data analysis dedicated to mining social media for telling information.

In 2012, for example, social analytics site Topsy Labs claimed it had developed a model that successfully predicted a drop in Netflix’s share price after the company decided to split its DVD rental and streaming video services.

Topsy did this by tracking negative phrases on Twitter like “just canceled my Netflix subscription” and found they outweighed positive phrases indicating customer satisfaction.

Then there’s big data company Wise Window, which operates under the tag line “Mass Opinion Business Intelligence.” The company’s stock model reportedly improved annual performance as much as 48% by tracking up to 2 million social media messages each month.

Unfortunately, these methods don’t yet work for individual investors. According to Neil McGovern of Skybase, the upfront cost is simply too expensive, and the lack of a proven track record makes it difficult to justify using such models.

Further, as has been the case with many forms of traditional quantitative investing, successful algorithms may not last forever. That is, once these models get into the hands of major hedge funds, the usefulness for retail investors at the back of the queue will diminish.

For now, it’s best for individual investors to stick with review sites, using them as supplemental sources of information.

Until next time,

Jason Stutman

Check us out on YouTube!