How to spot emerging technology innovation at public companies

How can you spot emerging technology innovation at public companies? And how could you scale this up, so that you could track not just a handful but hundreds or even thousands of companies?

But first, let me clarify what I mean by “emerging technologies, “innovation, and “public companies”.

By?“emerging technologies”?I mean technologies that are gaining in momentum — not necessarily brand-new technologies. For example,?digital twins?or?machine vision?are not new, but they have been gaining momentum over the past few years.

When I say?“innovation”, I mean something that’s more than just a paper, a patent, or a concept. There also has to be some tangible intent of business building. For example, a new product or project, a joint venture, or a new pilot plant.

By?“public companies”?I mean companies that are listed on at least one stock exchange.

And just to be clear, my goal here was not to do a full-blown, comprehensive analysis. Rather I wanted to test out a few methods that could be scaled up in the future.

My methods: The companies portfolio, data, and analysis

(If you’re not that interested in my methods, feel free to skip ahead to the results)

The companies portfolio

I made a portfolio of 126 companies from various industries. Some of the companies are household names such as BASF, Honeywell, or NVIDIA. Other companies are probably less well-known. In addition to companies I come across in my work, they include holdings of?ARK Invest ETFs, and some of the companies?Nanalyze?writes about.

I assigned each company to one industry sector, even though this isn’t always straightforward (more on this below).

Take a look at my my portfolio here ->

The data

I analyzed financial news, which is one of?Mergeflow’s data sets. The reason for using financial news is I was interested in technologies that already have some business footprint, as I mentioned above. And even with this limit, and a small portfolio of only 126 companies, my data set included more than 654,500 documents.

I did not consider patents and R&D publications because these are often not directly related to business building. In other words, there might be a patent or a paper but never any business that comes out of it.

The analysis

I wrote analysis code against?Mergeflow’s API. This allowed me to do bulk analyses, rather than just manual one-off searches. For my analyses, I did the following:

1. Get all financial news that mention any of my portfolio companies. I used financial news from the last 5 years. This way I got a data set of 654,500 documents.
2. In my data set, have Mergeflow identify any of?more than 200 emerging technologies?that are mentioned in the context of any of my portfolio companies.
3. Make 5-year, 1-month-resolution, time series for each “company x emerging technology” combination. In total, these were 126 companies x 228 emerging technologies = 28,728 time series.
4. Rank these “company x emerging tech” time series by “rising activity”: Those time series that had the highest “most recent 3 months” activity, compared to the overall time series, were ranked highest.
5. Using the ranking and some cutoffs (e.g. “number of news in last 3 months greater than 0”), I got a set of 18 highest-ranking “company x emerging tech” time series.
6. For the 18 highest-ranking “company x emerging tech” combinations, do some more content analyses to discover types of emerging tech innovation (more on this later).

Which companies show an increase in emerging technology innovation?

As mentioned above, I made 28,728 time series, one for each combination of?portfolio company?and?emerging technology.

I sorted the time series by “activity increase”. I defined “activity increase” as:

activity increase = average recent activity / average overall activity        

Then I applied various cutoffs, such as “time series must have more than 0 news during the most recent 3 months”, or “more than 5 news during last 5 years”.

After sorting and applying the cutoffs, I got the following 18 “winning combinations”:

When you look at this table, there are some combinations that are hardly surprising. For example, it’s quite obvious that “Web3” is rising at Coinbase and Ethereum, given what these companies do.

But there are combinations that I found more surprising. For example, “Dassault” and “in silico” and “bioinformatics”.

In order to understand better what’s behind these surprising combinations, I zoomed in on some of the combinations. I wanted to see if there are patterns or types of emerging tech innovation.

3 types of emerging technology innovation

Knowing more about types of emerging tech innovation would help make the approach here more scalable. For example, we could run analyses for much larger company portfolios, and try to detect innovation types.

But knowing more about innovation types might be interesting from a valuation perspective as well. Some types might be more valuable because they are likely to be more durable, for instance. Or there might be differences in risk; different types might be associated with different kinds of risk. I will come back to this below, after describing 3 emerging tech innovation types.

Why 3 types? It’s simple — I didn’t have the time to research more types. For the same reason, not enough time and not enough space here, I’ll only describe a few examples of each innovation type.

-> continue to the full article