Auquan Blog

A Quick Look At The 2020 Sentiment And Price Of IHG

July 25, 2020 by Chandini

In a previous post, we looked at an example of how extensive monitoring on top of comprehensive news coverage helps managers identify future risks and avoid significant losses. The challenge of scaling up your monitoring systems is two-fold. Firstly, there are just so many news sources out there that finding them all and collating their new releases is a never ending task. Secondly, even if you dedicate someone to solving this first problem, increasing the volume of data consumed leaves you overwhelmed and asking how to identify when information is actionable at scale?

It is clear that data must be accompanied by analysis to avoid information overload. In this article, we will explore how data science tools are uniquely solving these challenges. Using the example of International Hotels Group (IHG), we will explore how these tools allow you to monitor the complex web of connections that can affect a given company (both directly and indirectly) from an overwhelming sea of news, filings, research, market data and more.

IHG owns several brands, including InterContinental, Regent Hotels, Six Senses Hotels, Kimpton Hotels, Holiday Inn, Staybridge Suites etc across 100 countries. In order to truly cover all the news for a specific company, it is important to consider news about all these different entities. The way we do that is using a technology called knowledge graphs. This is a topic for another article, but here’s a snapshot of some of the most relevant entities for IHG:

In orange we have companies, blue are locations, brown are people, etc. The lines between circles are labelled with the nature of the connection e.g. ‘CEO’

To understand this company, you need to understand this entire ecosystem above (including the many parts excluded from the image). This means in any given week there will be hundreds of news stories about the company - just last week there were 300+ stories. Not all of them are relevant, or need attention, so the question becomes: How do you decide when stories are actionable?

Internally, we look at a few individual and aggregated metrics to signal times when information is actionable. A popular metric is the aggregate sentiment of news. While it may not be humanly possible to read individual stories, we are able to reliably extract text sentiment using automated processes. By looking at the trends in this aggregate sentiment (shown below) you can understand when the market mood about a name changes. Running this analysis for IHG, we see that the news sentiment suffered a huge dip, just before the broader market fell in February. Most investors were probably aware of rising concerns around travel, but the fall in sentiment would have highlighted how widespread those concerns were. The sentiment recovers in early March, as the US/UK maintain that they have the infections in control, before dropping again as most countries announce border closings and lockdowns.

If we accept that news is a proxy for average world opinion, then this sentiment change would have indicated changing market opinion well in advance. NB, It is important to ensure that we are measuring the sentiment towards the company, not the general sentiment of this news. Building on what we said earlier, a company is influenced by news outside of its immediate connections. For example, what is happening with its suppliers, clients, competitors, or policy changes in its countries of operation. Using our knowledge graph, we monitor these moves with a risk score that aggregates information and sentiment from news across a company’s universe. Note: Risk Score is a proprietary signal, but can be interpreted as the reflection of chance of a downwards price move (i.e. a higher risk score = greater chance of negative move). For IHG, we see that the overall risk score starts going up from early February, even before company sentiment fell. The risk score peaked in March, and has been steadily declining since, indicating that additional concerns over the hotel industry are shrinking.

It would be interesting to monitor how these metrics evolve over summer as European lockdowns have eased and travel has resumed. Again, we would be looking for signals to indicate the health of a company or sector ahead of time, before company trading updates are announced for Q2 and Q3.

Share this Article: