Four Ways to Understand Greek Debt
April 22, 2015Is the Raisin Reserve “Unraisonable”?
April 24, 2015Assume for a moment that a vast number of tweets convey some anxiety. We might conclude that on the previous day, the Dow had plunged. Yes?
Recent research suggests we reverse the sequence. Social media sentiment has correlated with a subsequent fall in financial markets.
Tweet Studies
In one study, the research targeted what they called emotional words that they then sought to correlate with financial market fluctuations. In the following table you can see the words they selected, the average number of times each one appeared in a tweet in one day (307 for “hope”), and a daily range (“hope’s” minimum was 54 and its max was 467).
More specifically, when “hope, fear and worry” predominated, the Dow dipped; when they subsided, the Dow rose. As a result, the study’s authors cited an inverse relationship between those emotions and the direction of the Dow.
In a similar study, looking at millions of tweets for emotional indicators, another research group concluded that a slew of calm tweets meant the Dow would probably rise. Anxious tweets and it fell. Their accuracy? An 86.7% success rate.
And yet another group sought to correlate social media sentiment on an hourly basis with market fluctuation. Their data indicated that 12 out of 48 financial instruments could indeed have some connection to social media sentiment.
While all of this research is pretty interesting, it left me with one question. Even if they are right–and the holder of the patent on a Twitter Predictor works–what about Flash Boys? Thinking of high frequency trading, Michael Lewis’s Flash Boys and what appears to be an algorithm revolution, I wonder if a sentiment index for social media targets the heartbeat of financial markets.
Our Bottom Line: The Wisdom of Crowds
Sentiment research on social media is about the crowd conveying more accurate information than individuals.
In The Wisdom of Crowds, New Yorker columnist James Surowiecki says that crowds can usually make decisions that are more accurate than individuals. In markets, crowds accurately price sodas and broccoli and tennis lessons. Studies demonstrate that crowds’ guesses cluster around the true number of jelly beans in those huge glass jugs. Similarly, bubbles are an example of crowd behavior. Here though, we have “collective decision-making gone wrong”.
Although we can debate the wisdom of financial sentiment research, tweet aggregates surely do provide crowd wisdom…or folly.