Social Market Analytics (SMA) creates actionable intelligence from social media data by filtering out the noise to deliver clean data on sentiment for financial markets. More specifically, it produces a family of metrics, called S-Factors, designed to capture the signature of financial markets sentiment.
Who uses SMA sentiment indicators? SMA data is suitable for all types of traders. High-frequency traders (HFT) use it as an input into technical factors. Quantitative traders use it as an additional factor in a multi-factor model. Risk departments also use it as an alert for predictive price and volume moves and Value at Risk (VAR) modeling.
Which markets work best? U.S. equities is the best-use case for SMA data. It is also used as a signal in U.S. indexes, forex, commodities and cryptocurrencies. Predictive signals range from a few minutes to quarterly depending on the model deployed.
How SMA Sentiment Works
SMA has a three-stage process to mine sentiment factors called (S-Factors) from Twitter (TWTR) and StockTwits. In Stage 1, the Extractor starts by extracting all signals for designated financial terms and symbols with no filter. In Stage 2, the Evaluator tweets and messages are analyzed for financial relevance using proprietary Natural Language Processing (NLP) machine learning algorithms; and in Stage 3, the Calculator determines the sentiment score for each stock using a sentiment dictionary adjusted for performance in the financial market domain (see “SMA intellectual property advantages,” below).
Methodology & Strategy
The strategy considers both S-Score sentiment and filters for the price of stock greater than $5 per share. The base strategy focuses on stocks with relevant positive and negative sentiment signals.
The strategy takes long positions in all stocks of the S&P 500 Index showing an S-Score higher than “2” (intermediate level of high positive sentiment) and short positions in stocks with average S-Score below “-2” (intermediate level of high negative sentiment).
These daily sentiment and volume factors are captured before the market opening at 9:10 a.m. ET and before the market close at 3:55 p.m., which allows the investor to take positions before the opening or closing of the market (see “Full history,” below).
The base strategy yields a Sharpe Ratio of 4.21 with a cumulated return of 654.47% during the period being analyzed, compared to a cumulated return of 53.10% for the S&P 500 benchmark, which yields a Sharpe Ratio of 0.79.
One final comment regarding this work’s conclusions has to do with the fact that length of history makes it difficult to backtest against different economic cycles. It should be researched further combining other factors to make the model more robust.
How much does it cost? How do clients receive data? The cost of SMA signal varies based on frequency and is a monthly subscription model. Front-end tools, APIs, and Excel Add-ins distribute S Factor Sentiment scores.