RavenPack uses 2iQ’s insider transaction data to analyze earnings calendar information
RavenPack is a leading provider of data analytics solutions to the financial industry. Leveraging nearly 20 years of research, it helps financial organizations generate actionable insights from unstructured data.
Recently, RavenPack utilized 2iQ Research’s insider transaction data in a research study focused on corporate earnings release date changes. The study used the RavenPack Earnings Dates data set, as well as the RavenPack Earnings Intelligence framework, which consists of earnings call transcripts, earnings-related news, and insider trading information, to explore how changes in earnings announcement dates can offer valuable insights about stock price movements surrounding earnings events.
Here’s a look at the study and how 2iQ’s insider transaction data was used in it.
Examining the impact of earnings release date changes
Before introducing the RavenPack Earnings Intelligence framework to the study, RavenPack’s data scientists examined insights contained in the RavenPack Earnings Dates data set. Developed in collaboration with Wall Street Horizon, this consists of earnings calendar change records for over 8,000 stocks globally since 2006, with a focus on the US market. Overall, the data set contains more than 242,000 earnings events recorded with unique event IDs and a full point-in-time history available for backtesting.
RavenPack started the study by exploring the differences in post-earnings price reactions between advanced and delayed earnings events within the US mid/large and small-cap universes. This involved constructing daily long-short equal-weighted portfolios within the US universe that were long the stocks advancing their earnings date and short those delaying their earnings announcement. Positions were initiated at the market close before an earnings release based on the advance/delay signal and exited at the following close, using close-to-close excess returns to analyze the strategy. Performance metrics were backtested across signal aggregation windows of one to 21 days.
RavenPack found that advancing/delaying earnings announcements can be predictive of positive/negative earnings results. The one-day aggregation window achieved the best results, with annualized returns of 8.0% for the mid/large caps and 19.7% for the small caps, and Information Ratios of 0.8 and 1.2, respectively.
Source: RavenPack
Next, RavenPack examined the price reactions around the actual changes in earnings announcement dates themselves by constructing daily long-short equal-weighted portfolios within the US universe but this time going long/short the stocks with advance/delay events on the close following the date change. Here, it found that the resulting portfolio strategy yielded annualized returns of 1.3% and 4.3% and Information Ratios of 0.4 and 0.7, for the mid/large and small caps, respectively, though mid/large caps performed better when using a 10-day aggregation window.
Interestingly, a blended approach combining the signals of the first two strategies enhanced overall performance. The plain combined strategy yielded annualized returns of 8.7% and 20.9% and Information Ratios of 0.9 and 1.4 for the mid/large and small caps, respectively.
RavenPack then introduced EDGE – which analyzes premium newswires such as Dow Jones Newswires, WSJ, Factiva, Benzinga, and other web publications – to incorporate the news sentiment associated with a company. The idea was that this might yield useful information about the market conditions at the time when an earnings calendar change event took place. Here, RavenPack devised a simple sentiment-enhanced strategy, conditioned on positive sentiment. For both the mid/large and small cap universes, positive sentiment conditioning resulted in better performance for the long-short strategies across all aggregation windows.
Incorporating the RavenPack Earnings Intelligence framework
RavenPack then took its analysis a step further by combining the earnings calendar changes with complementary earnings-specific signals. Here, it leveraged the RavenPack Earnings Intelligence framework, consisting of insights from earnings-related news, earnings call transcripts, and net insider transactions between reporting periods. These signals, together with the earnings calendar changes, cover different aspects of the business cycle and provide a broader, alternative view into the state of a company. To incorporate the insider transactions information, the signals were over or underweighted according to the net insider transaction volume executed between consecutive earnings announcements.
Using the RavenPack Earnings Intelligence framework, the data scientists constructed long-short dollar-neutral portfolios with a fixed size of 320 stocks, normalizing the long and short legs of the portfolio separately to reach 50% long and 50% short exposures.
The results were impressive. Combining Earnings Dates with the existing Earnings Intelligence signal resulted in a robust 15% increase in the Information Ratios for the US mid/large-cap portfolios over effective holding periods of a week or more. Annualized returns also improved, especially over shorter horizons, increasing from 15% to 18% for the shortest effective holding period.
In summary, RavenPack showed that advances/delays in earnings announcement dates can be predictive of positive/negative earnings results on the release date, and it was able to take advantage of this and build a profitable strategy around earnings announcement events. It also showed that by incorporating its Earnings Intelligence framework – which includes 2iQ’s insider transaction data – a consistent performance improvement was obtained.
Commenting on RavenPack’s research study, 2iQ’s Head of Data Science, Haris Chalvatzis, said:
“The value-add benefit of insider transaction data is the fact they can be used both as a standalone signal or enhance current signals. RavenPack study showcases how critical the insider information is, how it can be incorporate into an actual signal, and above all, how it can increase information ratio and overall portfolio performance."
Those interested in finding out more about how RavenPack used 2iQ’s insider transaction data to create profitable strategies around earnings announcement events can find additional information here.