A perfect predictor for stock market movements is the holy grail of investing. If you knew with certainty that the stock market would increase tomorrow, you would buy to take advantage of the change, and if yo knew with certainty it would decline, you would sell. If a predictor could be proven to be perfect, it would be its own undoing; if everyone uses the same decision making process and makes the same decisions, the predictor becomes the sole cause in this cause-effect relationships, and there would be no future movements to predict.
So predictors work best when not everyone jumps on board. But everyone wants this advantage. Mathematicians, engineers, and financial analysts try to come up with algorithms that will drive purchases and sales in order to make the most money, but perhaps social scientists have an advantage.
Researchers at Indiana University-Bloomington were looking at Twitter to see if language analysis could be used on the social network to gauge public mood. The assumption was that the public is generally happy after days where the stock market goes up and generally upset after days where the stock market tanks. They used the tumultuous recession in 2008 as the main data.
The results didn’t exactly meet the researchers’ expectations. Using standard word analysis used in medical research to determine the effect of pharmaceuticals, the moods measured on Twitter lined up with stock market movements — but three or four days in advance of those movements rather than as a result of the market activity.
According to the research, the mood on Twitter, qualified as “anxious” or “happy,” when added to a test algorithm based on stock market patters, improves the predictability of the stock market from 73.3 percent accuracy to 86.7 percent accuracy, a significant improvement.
This isn’t a perfect predictor, but it shows that public mood most likely has an influence on investors’ trades. It’s also important to point out that the public mood measured on Twitter doesn’t predict stock market activity alone, it’s only when this information is added to an existing moderately accurate pattern-based algorithm that Twitter moods are shown to make a difference.
There isn’t a causal relationship between moods on Twitter and stock market movements, of course, but if there is a correlation to be found, it could prove to be somewhat important, at least to fund managers looking for new ways to create and market equity funds. The same aspects of investor or public psychology that drive people to produce Twitter updates describing how they feel today could drive investors to buy and sell their investments.
A British hedge fund has already committed $40 million for investing using the researchers’ Twitter-based algorithms. So how has this fund, Derwent Capital Markets, performed so far? Its a hedge fund, so the information isn’t exactly public.
Market prognostication fascinates the mind of every investor. If there were just one way to know what stocks will do — and no one else were privy to this information — the lucky fortune-teller would never have to worry about money again.