Decreasing Average Communication in Social Media Predicts Downward Movements of the S&P 500.
- Sell signals in major downward phases of the S&P 500 index.
- Based on messages in social media about the index and its components.
- Data analysis period is longer than five years.
- Main sources are Twitter, Stocktwits, message boards, and regular news outlets.
- Average smoothed values based on total volume of messages have best predictive value.
- Aggregation level on an hourly basis. Buy and Sell signals are based on increasing and decreasing difference of aggregated smoothed reports.
- Model requires a threshold of message difference to trigger signals.
- Model has a strong long bias. Only 3% are short signals. Short signals, however, trigger timely in down phases of index.
- This leads to an outperformance of S&P 500 future index with a Sharpe ratio of 1.36 (S&P 500 at 0.94 in same period).
Dataset Featured in this Post:
US 500 Top Stocks
Blue Chip Companies