Kensho Technologies – Creating Stock Indices Using AI

With the coming Fourth Industrial Revolution, up to 80% of service jobs are expected to be impacted. We’ve identified 30 emerging technologies that investors need to be paying attention to if they expect to benefit from a technological revolution that few people see coming. While the most exciting emerging technology companies are private, there are still opportunities for retail investors in publicly traded stocks.

In the case of artificial intelligence as an example, there are just a small handful of stocks that would even come close to being considered “pure play”. On the other hand, robot stocks are plenty, though ETFs like the Robo Global Robotics & Automation Index ETF (NASDAQ:ROBO) also have lots of “non pure play” stocks. We like to use Motif Investing as a tool to create indices that we can use to track certain themes, like robotics as an example. Here’s our Nanalyze Robot Stocks motif which is essentially our own mini-ETF except that we don’t get charged any management fees:

(If you want to create mini-ETFs like the one seen above, sign up for an account with Motif Investing with no deposit required).

Here at Nanalyze we spend a great deal of time researching pure-play stocks for certain themes like nanotechnology, artificial intelligence, and robotics. Wouldn’t it be nice if we could use artificial intelligence to go out and find all the companies involved in the most exciting emerging technologies? One company called Kensho Technologies is using natural language processing to scour millions of pages of regulatory filings along with other unstructured big data sets to compile “new economy indices” as seen below:

Kensho Technologies Methodology

Before we go tucking into the Kensho indices, let’s take a closer look at who Kensho Technologies is.

About Kensho Technologies

Click for company website

Founded in 2013, Cambridge Massachusetts startup Kensho Technologies has taken in a total of $67 million in funding so far from a slew of big names like Citigroup, Google Ventures (their seed investor), Morgan Stanley, and Goldman Sachs which led their Series A back in 2014. It’s no surprise that Kensho technologies made the CB Insights AI 100 list with some of the high caliber people on their staff. Employees like one of the 7 members of the original iPhone engineering team, the creator of the world’s first solid-state two-bit quantum processor, and one of the youngest-ever entrants to Harvard will make you ponder why you’ve been such a slacker with your life so far.

Fast forward to this year and we see that in February, S&P Global led a $50 million Series B round which values Kensho at around $500 million. The fact that an index leader led the last round is quite telling as to what direction Kensho is taking. So what does Kensho do? In the broadest sense, Kensho addresses the 3 biggest challenges facing investment analysis on Wall Street today – speed, scale, and automation – by using natural language processing (a subset of AI).

Regular readers will recall a very interesting startup we highlighted before called Alphasense. Similar to Alphasense, Kensho uses natural language processing to query millions of documents like SEC filings to extrapolate useful information which can be queried using natural language. According to a recent article by Forbes, the Kensho Global Event Database is “one of the world’s largest repositories of data and information”. We’ve hypothesized before about building the World’s Smartest Smart Beta ETF, and this database is exactly what we had in mind for that task. However, in order to create an ETF you first need to create an index.

Creating Indices Using Artificial Intelligence

In a small corner of the Kensho Technologies website, you can find some very interesting collateral on a set of very cool looking indices that have been built using artificial intelligence. Kensho calls these the “New Economies Indices” and they use a different set of sectors instead of the traditional GICS sectors.

The fact that many emerging technologies span across multiple industries makes it difficult to classify stocks using a traditional industry hierarchy like GICS. In a previous article, we talked about how there is no “3D printing industry” just like there is no “artificial intelligence industry”. That is because these new technologies span across all the traditional industries we are familiar with so we need to create new industry definitions to accommodate them. That’s exactly what Kensho has done. They call these the “21st Century Sectors” and each of these sectors is an actual index itself as seen below:

Source: Kensho Index Methodology

Kensho then creates a universe of all the stocks found in the above indices and uses it to create their master index which they call the “Kensho New Economies Composite Index“. Interested to see what stocks are in this index? As most index providers do, Kensho will just give you the top holdings seen below:

If you want to see the entire list of 251 stocks in the index (82% of which are U.S. stocks), you’ll no doubt have to subscribe to it. As most index providers do, Kensho limits their indices to only include companies that have a market cap minimum of $100 million and a 3 month average daily traded value (ADTV) of $1 million. What is most important to note here is that the entire methodology is rules based. You don’t have to pay some portfolio manager for his genius stock picking capabilities so that he can underperform the broader market 80% of the time.

Now that the infrastructure is in place to create custom rules based indices using one of the world’s largest repositories of data, Kensho can take this in any number of directions. We still need to wait and see what sort of fees Kensho will charge for ETFs that are based on these indices. As a point of reference, the  Sprott BUZZ Social Media Insights ETF (BUZ) charges 75 bips which seems quite steep.

We’re stoked to see that a startup is actually using artificial intelligence for stock selection but question whether or not that many pure-play names exist at the moment to fill some of these niches. On the other hand, we’re humans so what do we know.

In future articles, we’re going to dig into each of these exciting Kensho Technologies indices to see what stocks they’ve picked and try to ascertain the “pure play” exposure we’re getting from each. While there don’t appear to be any ETFs tied to these indices yet, you can be sure that’s coming soon.

While NLP is now mainstream, there's plenty of growth left to be had. Four of the five AI stocks we're holding are SaaS businesses selling AI-powered enterprise software. Become a Nanalyze Premium annual member and get immediate access to our entire tech stock portfolio.

Leave a Reply

Your email address will not be published.