At Unicorn Bay, we designed an analytic engine using a set of newly developed Bayesian filters to perform opinion mining and analysis. Our engine monitors public channels in the news and social media 24 hours a day, 7 days a week to collect, curate and organize data, discovering the latest sentiments about your favorite stocks.
Our scanner gathers data on many popular stocks such as $AAPL (Apple), $FB (Facebook), $AMZN (Amazon) and many others. The mined data is then run through a semantic engine performing the actual emotional analysis.
The engine attempts to identify six basic emotions, namely: anger, sadness, joy, fear, disgust and surprise. Also, the engine performs a polarity analysis, determining the skew of the aggregate data towards a positive, negative or neutral emotional tone for that stock.
The aggregate data is filed and published to our users in a format that is easy to understand. We use Social Rate metric.
For example, we have an aggregate sentimental analysis chart for the first few weeks of May for Amazon (AMZN):
DOES SENTIMENTAL ANALYSIS WORK?
It works like a charm. Just compare the sentimental chart above with the actual chart of Amazon’s stock below:
- he slumps towards negative emotions on 5/15/15 is met with a similar drop in stock price on the same day.
- Notice the volume jump on that day and compare it to the sudden significant decrease on the sentimental chart. Sentimental analysis correctly predicted both the price and the volume. In this case, a sentimental analysis was the correct combined indicator of price and volume.
While not invincible, sentimental analysis has proven to be a useful (and usable) tool for predicting stock movements. The November 2013 $4.7 Blackberry buyout failure was first surfaced to the public through media, but before anybody even heard the news, Twitter feed sentiments predicted a drop in Blackberry’s stock.
According to social analytics company Dataminr, aggregated Twitter feeds can indeed demonstrate a high correlation with stock market returns.
Emotional analysis has real potential. Its use is growing in the trading arena.