Utilizing Social Media Sentiment for Market Insights
Context Analytics (CA) leads the industry in structuring and analyzing unstructured text for sentiment analysis in financial markets. One of CA’s flagship offerings, the S-Factor feed, delivers a comprehensive suite of social sentiment metrics sourced from Twitter and Stocktwits, differentiated by platform for greater accuracy.By grouping securities into sentiment-based quintiles, we reveal a strong correlation between online mood and next-day returns.
Data Sources and Quality Controls
On Twitter, Context Analytics uses a proprietary account rating algorithm to evaluate financial relevance and credibility of each account. Only accounts that meet the algorithm’s criteria are included in the S-Factor calculations. In contrast, all accounts on Stocktwits are included, as the platform is exclusively focused on financial conversations.
Mapping Messages to Securities
Each publicly traded security is associated with a tailored topic model—a set of rules and identifiers that links messages to specific companies. Messages that match these identifiers are ingested and analyzed for sentiment, which is scored on a scale from -1.0000 to 1.0000. These individual sentiment scores and message volumes are then aggregated over a 24-hour window and compared to a 20-day historical baseline, producing a variety of sentiment factors, or S-Factors. These metrics are updated every minute.
Understanding the S-Score
One of the most widely used S-Factors is the S-Score. It represents the exponentially weighted sentimentfor a security over the past 24 hours (S), relative to its 20-day historical mean (s-mean) and 20-day historical standard deviation (s-volatility). This approach weights recent messages more heavily than older ones, capturing shifts in sentiment in near real-time. Because the metric is normalized to each security’s historical behavior, it avoids bias from message volume alone.
Research: Combining Twitter and Stocktwits Sentiment
We explored the predictive power of a blended S-Score by averaging Twitter and Stocktwits S-Scores for each security at the specified timestamps. Only securities with sentiment activity on both platforms and a share price above $5 at the previous day’s close were included. Each day, securities were sorted into quintiles based on their average S-Score prior to market close (3:40pm ET):
- Quintile 5: Top 20% (most positive sentiment)
- Quintile 1: Bottom 20% (most negative sentiment)
We then calculated daily Close-to-Close returns and cumulative returns for each quintile.
Top Quintile vs. Bottom Quintile:
- Cumulative Return: 83% vs. 12.49%
- Sharpe Ratio: 0.65 vs. 0.22
- Average S-Score: 974 vs. -1.115
- Return Spread +79.32% return
Key Findings: Sentiment Predicts Returns
The results clearly show a positive correlation between average S-Score and subsequent daily returns. Securities in the top sentiment quintile consistently outperformed those in the lower quintiles. Additionally, the bottom quintile consistently underperformed its peers.
The S-Factor feed is Context Analytics’ most established product, backed by over 10 years of historical data and a suite of 15 distinct sentiment factors. This rich dataset enables comprehensive back testing and empowers users to integrate real-time social media sentiment into financial strategies.
Interested in learning more? Reach out to us at ContactUs@ContextAnalytics-AI.com or visit www.contextanalytics-ai.com for more information.
TL;DR: Context Analytics uses social media sentiment from Twitter and Stocktwits to predict stock performance. Stocks with the highest average sentiment (S-Score) returned nearly 9x more than those with the lowest, proving that real-time online sentiment can forecast short-term market movements.
Strategy Summary: In review of the strategy and the key findings, we can summarize the results as follows:
- Sentiment Source:Twitter + Stocktwits
- Grouping:Stocks sorted into 5 quintiles by sentiment
- Top vs. Bottom Performance:
- Cumulative Return:Q5 = +111.83%, Q1 = +12.49%
- Annual Return:Q5 = +16.95%, Q1 = +6.14%
- Sharpe Ratio:Q5 = 0.65, Q1 = 0.22
- Sortino Ratio:Q5 = 1.04, Q1 = 0.34
- S-Score:Q5 = 1.974, Q1 = -1.115
Conclusion: Higher social sentiment correlates with stronger next-day stock performance.