Quandl goes live with new dataset for measuring dollar value of patents

Nasdaq-owned alternative data provider Quandl is live with a new dataset meant to estimate the value of companies’ intellectual property and patents, a growing area of interest among consumers of alternative data.

Assets such as IP rights and trade secrets are intangible assets—non-physical but valuable items not included in balance sheets. According to a May 2019 report by European alt data vendor Neudata, intangibles rose from 17% in 1975 to more than 84% as a percentage of total assets on S&P 500 constituents’ balance sheets. “Many datasets that measure intangibles are relatively unknown and under-utilized,” the report reads.

Although patent data is publicly available from sources like the US Patent Office, it can be difficult to structure and analyze properly. On top of that, it is updated infrequently. But patent and IP data, which are increasingly being examined by quant funds, can offer a key to separating truly innovative companies from those merely paying lip service to innovation. However, Bill Dague, head of alternative data research at Nasdaq and Quandl, hopes the new offering will provide alpha advantages to both quantitative and fundamental investors due to its wide coverage, monthly updates, and proprietary evaluation methodology.

Quandl has partnered with a global consultant, which it can’t yet name, that works with businesses on developing, protecting, and valuing their IP. Quandl has used statistical analysis to quantify the partner’s methodology, and the two companies built a proprietary algorithm to apply systematically to companies’ patents. The algo is meant to determine the relative value of patents within an industry or sector, and scale that value by the size of the company and the impact that similar patents have had.

After applying this methodology to the raw data, Quandl maps it to the parent entity’s ticker, creates a baseline against its own indicators, and comes up with a dollar estimate of the potential revenue impact in the medium to long term. The dataset covers 20,000 tickers globally, leveraging 10 years of patent history, and includes even defunct companies for comparison purposes.

Using and evaluating this type of data can be challenging, as there is a lot that can confound the analysis. For example, the interpretation of “innovation” is subjective. The fact that a company has a lot of patents, for instance, doesn’t necessarily indicate its actual performance as an innovator, says Octavio Marenzi, CEO and founder of capital markets consultancy Opimas. Depending on the industry the company operates in, a company’s IP and patents can be a very important indicator of its value. But this particular insight, as a subset of alternative data, has been slow to emerge mostly because it isn’t easy to extract data out of a patent, he says.

“Let’s say you have the patent for the pencil: It would be a simple one, and it would be worth a fortune,” Marenzi says. “However, what data in the patent would allow you to determine this?”

Bringing algorithms and patents together to automate this process strikes him as “fiendishly difficult.” 

“It’s one thing to have a robot read a million patents, but then how to make sense of the information is far from obvious. Most patents are written in rather free form text and even if you can convert it into some approximating structured data, figuring out what a patent is worth is another enormous step. Two quite similar sounding patents might have entirely different values, once you consider the context of the industries in which they operate,” Marenzi says.

Indeed, the development of Quandl’s dataset has involved some trial and error. It had experimented with IP datasets before, Dague says, having originally used a ranking system as opposed to the quantified, dollar-amount approach. But it didn’t add much value, and the company hadn’t fine-tuned its understanding of patents’ nuances—identifying patent harvesters, sifting through legal language, and the room for interpretation that the word “innovative” can leave.

“There’s all sorts of noise in terms of how a patent manifests in the real world,” he adds. 

Tech giant Apple, for example, owns a patent for a circular pizza box design with air holes meant to reduce crust sogginess, which it invented for its headquarters’ cafeteria. “Is Apple going to make more money because of that patent? Probably not. Some companies have a strategy of just filing tons and tons of patents, or patent harvesting—buying up other companies to get patents,” Dague says. “Does that make them more innovative?” 

Even so, Dague says, in back-tests of the new dataset, Quandl conducted a rebalancing strategy on the Stoxx Europe 600 index, and found that in aggregate over the long term, the true “innovators,” based on evaluated patents, over-performed.

The patent value estimates could also be useful in the context of Covid-19, as the alt data market shows interest in data on industries like airlines, hospitality, medical stocks, and supply chains, all of which, Dague says, remain in high data demand by clients. In the Covid-19 era, growth stocks have outperformed value stocks over and over. Investors hope that IP and patent data could help them spot the next generation of money-makers and unicorns as the world inches toward a post-pandemic future.