The Power of Combining News and Social Media: 2025 Performance Update

In 2024, we demonstrated how combining sentiment signals from Twitter, Stocktwits, and News could enhance trading performance by capturing multiple, uncorrelated perspectives on market mood. Since then, we’ve continued to refine this multi-source sentiment framework, and the results from 2025 reinforce the power of diversification across data sources.

Persistent Low Correlation Across Sources

Our earlier research showed a Pearson correlation below 0.3 between the sentiment metrics derived from Twitter, Stocktwits, and News. This year, we expanded that analysis to a time-series view of rolling correlations, confirming that these relationships remain consistently low over time.

Even during volatile market periods—such as the election cycle and tariff uncertainty—the three sources continued to provide distinct sentiment signals.

• Twitter vs. Stocktwits: modestly correlated but dynamic, reflecting differences in retail investor communities.
• Twitter vs. News: near zero correlation, as real-time crowd emotion diverged from the tone of professional journalism.
• Stocktwits vs. News: similarly uncorrelated, providing an independent lens on sentiment momentum.

This stability in low correlation underscores the structural advantage of multi-source modeling—each platform captures a different dimension of investor behavior and narrative flow.

2025 Year-to-Date Portfolio Performance

Building on our 2024 framework, we continue to trade daily close-to-close rebalanced quintile portfolios across U.S. equities priced above $5. The performance through 2025 has been exceptional:

• Long “All Sources” Portfolio: +33% year-to-date, outperforming the SPY by over 16%.
  • Sharpe Ratio: >1
  • Sortino Ratio: >2
• Short “All Sources” Portfolio: –20% year-to-date, confirming the strength of our bearish sentiment identification.

When combined into a theoretical long–short portfolio, the cumulative performance would exceed +50% for the year. These results validate that multi-source consensus—where all three signals align in the same quintile—continues to produce alpha even in changing market regimes.

Why It Works

The resilience of this strategy lies in signal orthogonality and timing diversity:

• Social Media captures immediate crowd reaction and speculative flow.
• News reflects institutional tone and broader macro context.
• Cross-source consensus identifies moments when retail and professional sentiment converge—often precursors to outsized price movement.

By weighting and aggregating these distinct sentiment signals, Context Analytics clients gain a more holistic, low-noise sentiment index that enhances both long and short opportunities.

Conclusion

The 2025 results reaffirm what our original analysis revealed: combining uncorrelated sentiment sources amplifies predictive power and reduces noise. At Context Analytics, we remain committed to transforming unstructured text into structured, actionable data that drives investment performance.

For more information, visit www.contextanalytics-ai.com.