Does sentiment on StockTwits and Twitter social media platforms usefully predict returns for individual stocks? In their June 2018 paper entitled “Momentum, Mean-Reversion and Social Media: Evidence from StockTwits and Twitter”, Shreyash Argarwal, Pablo Azar, Andrew Lo and Taranjit Singh analyze relationships between stock price behaviors and real-time measures of sentiment uniquely attributable to StockTwits and Twitter in three ways:
- Linear regressions for a sample of 4,544 stocks that each day relate volume and liquidity metrics for each stock to aggregate news and social media sentiments for that stock measured either during the same trading day (9:30AM to 4:00PM, for coincident relationships) or during preceding non-trading hours (4:00AM to 9:30AM, for predictive relationships).
- An intraday event study for a subsample of 500 large-capitalization stocks that examines stock trading behaviors when associated bullish and bearish social media sentiment reaches extreme levels.
- A backtest of an intraday mean reversion strategy applied to the 500 companies with the highest average volumes over the previous 200 days (with no more than 30% from a single sector) that exploits the power of social media sentiment to predict mean reversion. Every 30 minutes, this strategy buys (sells) stocks with negative (positive) returns over the preceding 30 minutes, with weights elevated for stocks with high StockTwits and Twitter message volume over the preceding 30 minutes.
Using the RavenPack Composite Sentiment Score to measure conventional stock sentiment, minute-by-minute StockTwits and Twitter-with-retweets data from PsychSignal to measure social media sentiment, and trade/quote data for 4,544 stocks during 2011 through 2014, they find that:
- Based on regression tests:
- There is much more demand for and much less supply of liquidity for a stock when its social-media sentiment is negative than when it is positive. Conventional news sentiment cannot explain this relationship.
- Negative social media sentiment has a much larger effect on liquidity than positive sentiment. A 1% increase in bearish sentiment has twice the impact of a 1% increase in bullish sentiment.
- Pre-trading hours measures of both conventional news sentiment and social media sentiment predict liquidity during the ensuing trading day, with high sentiment preceding high liquidity demand and a relatively high probably of mini-crash.
- The event study shows that abnormal social media sentiment (at least three standard deviations above or below average) for a stock follows strong momentum and precedes mean reversion of returns.
- The intraday mean reversion strategy that does (does not) adjust position weights using social media sentiment generates gross annualized return 24.1% (20.6%). Restricting trades such that they never require more than 10% of available volume for a stock at any second reduces this return to 14.8% (12.6%). However, such high-frequency trading is costly, such that the strategies would be profitable only for market makers with very low trading costs.
In summary, evidence suggests that social media sentiment may help very low-cost traders exploit mean reversion of stock returns by identifying stocks most likely to turn up or down.
Cautions regarding findings include:
- As noted, backtest performance results are gross, not net, and trading frequency would drive high trading frictions.
- Annualized return is not sufficient to determine strategy attractiveness. A high-volatility strategy may entail ruinous drawdowns.
- The services required for collecting and processing sentiment data may be costly.
- Given the rapid evolution of social media, a sample ending in 2014 may be stale. In other words, automated traders may have exploited and lowered reported gross profitability.
