[ibimage==39317==Large==none==self==ibimage_align-center]
Good investment often means knowing more than the market and banking on that knowledge imbalance. In fact, Warren Buffet’s style of investing, value investing, is just that: investing in stocks the market is unreasonably discounting in the belief that market knowledge will eventually catch up and boost the price. Chicago startup Social Market Analytics has taken that philosophy heart and is using sophisticated software to measure market knowledge via Twitter. Using that data it alerts investors to potential investment opportunities(and investment dangers). Since the company just raised over $750,000, Built In Chicago caught up with Social Market Analytics' co-founder and CEO Joe Gits to hear about this innovative use for social data:
[ibimage==39308==Large==none==self==ibimage_align-left]
How can people profit from better understanding the public's sentiments about stocks on Twitter?
Twitter is becoming more and more important as a means of distributing news. As an example; last year the SEC said companies can meet their corporate disclosure requirements by posting on Twitter. By taking a continuous feed of Twitter, filtering/cleaning and creating metrics we are able to glean the intentions of professional investors. These intentions lead to subsequent movements in stock prices and volatility at a statistically significant level.
How do you differentiate negative sentiments from positive sentiments?
We capture the intentions of professional investors. To identify those professional investors we have developed 17 metrics. Once we identify those investors we look at the Tweets and identify those Tweets that discuss their intentions and opinions. We parse the Tweets using Natural Language Processing (NLP). We look at phrases and words and score each Tweet based on its composition. Positive and negative Tweets are identified by the words and phrases used. We give each Tweet a score. Just identifying a Tweet as positive or negative doesn’t provide context or strength of opinion. By fine grain scoring we identify the level of positive or negative.
When someone sees an advantageous stock score on your platform, typically what is their time window to act (before market activity catches up with Twitter sentiment)?
It really depends on the market cap of the security. For the largest most active names (AAPL, GOOG, FB, MSFT, IBM) the time to act is short because the traditional news media follows those stocks in real-time. For the rest of the rest of the universe you have a good amount of time. Our clients have told us the signal persists up to a week out.
Any other new products or initiatives that you are excited about?
We are really excited about some of the new products we will be releasing this year. Commodities and currencies will be released shortly. We just announced a partnership with StockTwits (the number one social media site for finance) to provide our proprietary metrics on their social conversations and we will start providing real-time publication of scores on individual Tweets through our API.
How does the platform give people context?
We measure the level of agitation in social media relative to historical levels for individual stocks. Sector, industry, and index sentiment scores are calculated as well. We describe the current social media conversation relative to a standard normal curve. If an S-Score is 2 the current conversation is 2 standard deviations more positive than prior conversations. That means the current conversation is more positive than 96% of prior conversations. S-Delta is the change in S-Score and S-Dispersion is a proprietary measure of the breadth of the conversation.
- S-Score™: is the weighted normalized representation of our sentiment time series over a pre-defined look back period.
- S-Mean™: is the weighted average of the sentiment scores on a given day, over a pre-defined look back period.
- S-Delta™: is the percent change in S-Score™ over a time period. It is a first order measurement of the sentiment trend.
- S-Volume™: is a measurement of the volume of indicative tweets used to calculate the S-Score™.
- S-VScore ™: is the weighted normalized representation of indicative tweet volume over a pre-defined look back period.
- S-Volatility™: is a percent measurement of the variability of the sentiment level over a pre-defined time period.
- S-Dispersion™: is a measurement of the tweets concentration factor responsible for the S-Score™.
