<img height="1" width="1" style="display:none;" alt="" src="https://dc.ads.linkedin.com/collect/?pid=586106&amp;fmt=gif">

SUBSCRIBE HERE


Posted by Daniel Krastev
Daniel Krastev
Find me on:

Ambiguity bias – Avoiding the unknown. We all exhibit biased behavior – even when we think otherwise. Growing up, most of us are conditioned to distrust the unfamiliar. Think back to a time when you’ve bought a brand name product instead of its generic brand counterpart. It’s highly likely that you paid substantially more for the brand name product. The surprising fact is that many generic brand products are manufactured in the same facility, using the same raw materials, as the brand name product - the only difference is the packaging and the price.

In the finance and capital markets fields, M&A analysts have traditionally relied on vetted sources, such as the Wall Street Journal, to identify potential M&A market chatter. Traditional paradigms, however, have become ineffective in the modern digital age, when unstructured data has grown exponentially. Today, the number of rumors published by low-popularity M&A news sources, such as Twitter, greatly outnumbers that of high-popularity sources. Therefore, it’s possible that M&A analysts’ ambiguity bias is preventing them from ingesting a majority of the informational alpha in M&A market chatter.

Is there informational alpha in low-popularity M&A sources?

The reason why high-popularity sources are believed to have high informational alpha is because of their reach and vetting process. In order to explore the relationship between the popularity of a source and the alpha content of the rumors they disseminate, we will analyze 3,000 M&A rumors that Rumor Hound has captured by crawling 100,000 unique sources. Our analysis will focus on answering the following questions:

 

  • Do rumors published by low-popularity sources influence the stock price of M&A targets? How does that impact, if any, compare to that of high-popularity sources?
  • Do low-popularity sources exhibit a faster response-time than high-popularity sources?
  • Do low-popularity sources have a significant affirmation rate (are they frequently followed by high-popularity sources)?

 

If our research findings positively affirm these questions, we can conclude that low-popularity sources contain enough informational alpha to tip the scales in their favor.

Download the full research paper right now and discover our findings and what conclusion we draw

Massaging the data – Filtering out noisy market chatter

Some rumors do not seem to exhibit a significant impact on general market sentiment. These rumors are typically ones that are disseminated by just one or two sources and then fade out. Conversely, some rumors are disseminated by numerous sources, some of which are highly—popular, widespread, and credible. So, before we start our comparative analysis, it is necessary that we define a “significant event.” A significant event is defined as:

 

  • A series of at least four individual rumors published by unique sources, where
  • At least one rumor is published by a high-popularity source (i.e. Street Insider, Wall Street Journal, etc.)

The initial step in our research will be to filter our dataset for significant events.

Exploratory analytics – Looking inside the rumor

Our aim is to expose some descriptive statistics regarding significant rumor events. Our analysis will include:

 

  • Plotting the distribution of rumors sourced by low-popularity and high-popularity sources. This will show the relative prevalence of hidden, low-popularity rumors to high-popularity rumors.
  • Filtering the significant events into two subsets: 
  1. For significant events started by a low-popularity source, find:
    • Average number of high-popularity followers.
    • Average number of low-popularity followers.
    • Distribution of high-popularity and low-popularity sources.
    • Mean, variance, skewness, and kurtosis of percent price impact.
  2. For significant events started by a high-popularity source, find:
    • Average number of high-popularity followers.
    • Average number of low-popularity followers.
    • Distribution of high-popularity and low-popularity sources.
    • Mean, variance, skewness, and kurtosis of percent price impact.

 

We will use these findings to compare the informational alpha content of low and high-popularity sources.

Why is this important from a business standpoint?

“We cannot change the cards we are dealt, just how we play the hand.”

-Randy Pausch, The Last Lecture

The quality of a strategy is only as effective as the factual information and assumptions upon which it is built. In order to devise effective strategies, M&A analysts, bankers, and traders need to be armed with factually-reliable content, at the right time. Given the explosion of unstructured data, it’s becoming increasingly difficult and time-consuming to find reliable data at early timeframes. If low-popularity sources do, in fact, contain informational alpha at earlier timeframes, then M&A professionals can leverage this information to generate alpha more effectively, make smarter adjustments on short positions, mitigate risk exposure, etc. Stay tuned for our research findings and discover if unpopular and inconspicuous M&A news sources contain alpha that you may be missing out on.

Get research paper


Topics: Artificial Intelligence