At Context Analytics, we specialize in transforming raw data into actionable insights that can inform investment decisions. One of our key offerings, the S-Factor feed, derived from Twitter conversation, has been a cornerstone of our approach for years.
S-Factor Feed
The S-Factor feed leverages real-time data from Twitter to generate trading signals, dashboards, and widgets that help investors enhance their strategies. By analyzing the vast volume of tweets, our feed provides insights into market sentiment, allowing investors to make more informed decisions based on current trends.
A unique feature of the S-Factor feed is its flexibility—investors can manipulate the metrics to create customized metrics that align with their specific investment goals. This blog explores one such innovation: a strategy that uses a metric derived by multiplying the S-Score and S-Buzz to optimize portfolio returns over a multi-day holding period.
Developing a New Metric: S-Score * S-Buzz
The S-Score represents normalized sentiment over the previous 24 hours, while S-Buzz measures unusual tweet volume relative to the broader market. By combining these two metrics, we enhance the impact of sentiment signals when there is heightened tweet activity, making it a useful tool for identifying trading opportunities. We developed a long-short trading strategy that uses a metric created by multiplying the S-Score and S-Buzz. This combined metric reflects both sentiment and volume, providing a reliable indicator of market momentum and helping us identify entry and exit points for trades.
For long positions, the strategy triggers an entry when this metric rises above a threshold of three standard deviations above its mean. This threshold indicates an exceptionally positive sentiment and unusually high volume, suggesting a strong upward momentum. The long position is maintained as long as the metric stays higher than one standard deviation above the mean, which acts as a trailing stop. This exit criterion ensures that the position is held until the sentiment and volume indicators begin to normalize, thus capturing as much of the upside as possible while managing risk.
Conversely, the short strategy is designed to capitalize on negative market sentiment. It mirrors the long strategy but focuses on signals where negative sentiment is coupled with increasing volume. An entry is triggered when the metric drops three standard deviations below the mean, signaling extreme bearish conditions. The short position is then held until the metric recovers to one standard deviation below the mean, allowing us to ride the downward trend while controlling risk as the market begins to stabilize.
This strategy assumes that the distribution of our metric holds consistently over historical data, enabling us to use these static thresholds confidently. By relying on this historical distribution, we can maintain consistent entry and exit points that are both statistically robust and reliable across different market conditions.
The following visualization demonstrates the distribution of this metric along with the specific thresholds that trigger our entry and exit signals for both long and short positions.
Strategy Performance
We applied this strategy to a universe of stocks with prices above $5, and using the Russell 3000 as our benchmark. The strategy involves trading from close to close, with daily compounding and rebalancing.
Since 2016, the long strategy has outperformed the Russell 3000, enhancing risk-adjusted returns. While the short strategy underperforms on its own, it effectively complements the long positions in a combined long/short approach, delivering an annual outperformance of 9% over the Russell 3000. The Long/Short strategy also achieves a higher Sharpe and Sortino ratio compared to the benchmark. Long positions have an average holding period of 1.32 days, while the short positions last longer with its average holding period over 2 days.
Conclusion
This example highlights the versatility of Context Analytics’ S-Factor data in enhancing trading strategies. By refining existing metrics, we've demonstrated how you can generate substantial returns over multi-day holding periods, offering an additional method to leverage Twitter data for portfolio management. For more information on the S-Factor feed or to start a trial, visit www.contextanalytics-ai.com .