PsychSignal Case Studies
Statistical Dynamics of Social Media Mood Data & Intra-Day Volatility
In this paper we provide empirical evidence that PsychSignal's Natural Language Processing System can predict an asset’s future price volatility. We use the NLP to scan the social media fabric for anomalous behavior from traders and investors. Our research suggests a causal link between social anomalies and future stock price behavior.
We conducted several experiments to measure how effective short term Cash-Tag Volume Anomaly Scores (“CVAS”) can predict volatility. In addition, we measured the effectiveness of the medium term CVAS in predicting daily volatility for a given asset. In our experiments we use three ETFs: (SPY, DIA and QQQ) as well as the constituents of the Dow 30. The results serve as guidance on how to intuitively and practically apply a CVAS to trading models and systems.
“Pre-Market anomalous sentiment activity leads increased market price volatility”
Measuring the Effect of Short Term Social Sentiment on Stock Market Correlation Structure
We studied the effect of Short Term Social Sentiment on intraday correlations between individual stock symbols and the S&P 500 (as measured using SPY as a proxy). The data suggests a strong relationship between the current level of social activity and future price correlation on an intraday basis.
Portfolio optimization is based on the effect of diversification between a group of assets; undetected changes in correlation can thus expose market practitioners to unwanted price risk. This implies substantial benefits to incorporating social data into risk management and trading applications.
“The results imply that when social activity is high in $SPY, the correlation of $SPY increases with a broad range of individual stock symbols and ETF’s.”
How sentiment data Improves Absolute Momentum Strategies for the S&P 5oo
In this paper we show how incorporating a socially derived dataset can produce an augmented absolute momentum strategy for trading the S&P 500 index (through the SPY ETF as a proxy).
This augmented strategy achieves: Superior Sharpe Ratios, Higher win percentages and ending equity, Less severe max drawdowns and Spends less time invested in equities.
“...including a social activity filter significantly improves the risk adjusted returns of
absolute momentum strategies.”
PROFOUND IMPACT OF SOCIAL SENTIMENT ON PAIRWISE CORRELATION STRUCTURE
We find that future intraday pair-wise correlations are profoundly affected by social media activity. Specifically, this paper describes the existence of a triangular relationship between social activity and correlation whereby the social activity of the $SPY exchange traded fund (ETF) acts as a statistical switch, influencing the relationship between the pair’s correlation and their individual levels of social activity.
“Using social media data-streams, such as the PsychSignal Trader Mood data, can significantly improve our understanding of the time evolution of correlation.”