Innovative businesses are more likely to outperform regressive ones in the long run—this assumption is something akin to gospel for companies and investors alike. 

But a problem arises when investors try to concretely account for the innovativeness of a particular company: innovation is intangible and thus difficult to evaluate. 

Thankfully, there is an alternative data answer to the problem.

Patents are an excellent way to evaluate a company’s innovativeness and uncover value that doesn’t show up on balance sheets. They are a powerful leading indicator and can signal a company’s potential future growth. Developing new products, new technologies and licenses—and patenting them, of course—can bring in stable profits and help a company edge out lagging competitors.

A next-generation approach to measuring innovation

At one point in time, quantifying and measuring a company’s innovativeness via patents entailed obtaining a list of patents from public patent information sources. One would then do a simple count of the patents to ascertain a company’s level of innovation.

There is some legitimacy in this approach. A firm with 100 patents is more likely to be innovative than a firm with just one patent. But the approach lacks nuance and is limited in how much actionable information it can be expected to provide. What if the single patent is extremely valuable, while the 100 patents are little better than junk?

Nasdaq’s Quandl is now offering a more insightful, sophisticated and nuanced approach to patent valuation with one of our latest products, Patent Value Estimates—the next generation of patent valuation. 

The data set covers over 20,000 active and delisted companies in over 100 countries, making its wide coverage more useful to those investing internationally and across sectors.

The product uses a proprietary valuation methodology to arrive at actual dollar estimates for a large universe of patents. The result is more accurate data for quantitative investors and a more tangible understanding of a company’s intellectual property assets for non-quantitative investors.

The process of patent valuation

By combining multiple techniques for patent valuation—referencing patent value portfolios, digging into financial statements and benchmarking patents against each other—our data scientists are able to assign an actual dollar value for a large number of patents. 

The first two approaches are accessible enough for a determined data science team, but they cover only a small number of patents. The act of benchmarking patents against each other is the real game-changer when it comes to Patent Value Estimates. 

The approach involves using artificial intelligence modelling of patent filings to capture similarities between different patents. If a patent was sold or bought and held a certain dollar value and there exists another patent very similar to that one, it’s reasonable to assume that the second patent’s value is comparable to that of the first. 

This adds a critical dimension to the proprietary patent valuation methodology, which in combination with the other two approaches, enables us to provide actual dollar estimates for a large universe of patents. 

Taking the data for a test run

The Patent Value Estimates dataset is more than just clever data science—we’ve also run a rigorous backtest to ensure that patent value is indeed an indicator of future performance.

We began with the thesis that, for companies with higher patent value, one would see higher performance of their stocks in the long term. 

Naturally, there was also normalization to be done at the outset. For example, it would be unfair to compare small firms’ patent value to larger firms. After all, the Googles and Microsofts of the world have larger R&D resources on hand than most companies. To account for this, we normalized a company’s patent value by dividing it by the firm’s total assets. 

Our testing pattern was to create a patent value factor, which is the ratio of each company’s estimated pattern value to the total assets. If the ratio is lower, we short the stock and if it is higher, we long the stock, forming a long-short backtesting strategy. 

Testing against the STOXX Europe 600 by buying companies with relatively large patent value factors and selling companies with relatively small patent value factors delivers a Sharpe ratio 50% higher than the underlying index. Further, the strategy remained performant under a variety of robustness checks. (You can find a more detailed report on our backtesting here.) 

A leading indicator of success

A tangible, accurate measure of a company’s intellectual property is a powerful addition to both quantitative and non-quantitative strategies. As a leading indicator of future success, patent value estimates can help investors decide the value of a stock and augment their strategies with more relevant and accurate information.