Enhancing Short Term Stock Reversal Strategies with News Analytics

Executive Summary:

Here is a summary of our findings about short term stock reversal strategies :

• we find stronger reversal effects for past losers when unsupported by news.
• We also find strong reversal effects for past winners when supported by news.
• Perhaps unsurprisingly, reversal effects are stronger for past losers when company sentiment is positive and for past winners when company sentiment is negative.

This study examines all S&P 500 constituents using a 1-week investment horizon.

News Analytics significantly improves returns

• Buying past losers without news volume and selling past winners with negative sentiment generates an annualized return of 7.3% with an Information Ratio of 1.47 after transaction costs
• Performance is further improved filtering buy signals on low price volatility and filtering sell signals with high price volatility. In this case, the strategy generates an annualized return of 14.6% with an Information Ratio of 1.54 after transaction costs.
• Overall, we find significant alpha across all models and beyond the benchmark reversal signal – something which cannot be explained by traditional risk factors.

Introduction


Short-term reversal is a well-established phenomenon in financial markets. On shorter time horizons, such as a week or a month, stocks tend to mean-revert on average. Buying past short-term losers and selling winners typically yields positive excess returns over time. In the literature there are two explanations for short-term reversal profits: (1)an overreaction caused by sentiment; or (2) one caused by trading friction.

The sentiment-based theory argues that market prices may reflect investor overreaction to information, or fads, or simply cognitive errors. The trading friction theory, on the other hand, suggests that the reversal in price is due to the temporary price concession required by the liquidity providers as compensation for uninformed trades.

Due to the difficulty in obtaining a direct measure of sentiment around a given company, only a few empirical studies have performed tests of the sentiment-based theory on short-term reversal (e.g. Liu and Schaumburg, 2011). But with RavenPack News Analytics, a real–time news analytics service that covers more than 31,000 publicly traded companies worldwide, it’s now possible to quantitatively construct news-based indicators at the company level (in real time) and thereby obtain a better understanding of the role of company sentiment and news volume in the pricing of individual assets.

Based on the RavenPack News Analytics dataset, we construct two proxies for sentiment including: (1) an event volume indicator capturing the count of news events per company, and (2) a company sentiment indicator that considers RavenPack’s event sentiment score (ESS). We assume companies have “news” when they experience high news event volume and “no news” when it’s low or there are no relevant news events covered in the media. We find strong evidence that news analytics significantly affects short-term reversal. Specifically, we find (a) strong reversal effects for past losers with low event volume and for past winners with high event volume; and (b) stronger reversal for past losers with positive news sentiment and for past winners with negative news sentiment.

In the following section, we provide a brief description of our two news-based indicators. In Section 3, we evaluate a simple reversal strategy with a news analytics overlay. In Section 4, we further investigate the impact of high vs. low stock volatility on the news analytics enhanced reversal strategies. Section 5 examines the exposure of the news analytics overlay to the Fama-French risk factors. Finally, in Section 6 we present our conclusions.