AI Analyst™ Terminal

Artificial Intelligence based IP Analytics. Smart Beta for your portfolio. Find matching M&A candidates. tracks more than 110 million patents, 220 million scholarly articles and 115 million source code repositories worldwide - basically almost all patents, scholarly work and open source codes in the world - and provides AI based analytics of Intellectual Property, matched with public and private companies from more than 70 countries. With the AI Analyst™ Terminal, asset managers can generate unique smart beta for their investment decisions or M&A advisors can auto-generate IP matched short-lists. Access to the platform is strictly limited.

  • Smart Beta for your portfolio
  • Auto-generate M&A short-lists
  • Integrated IP indicators
  • Stock Details and Indicators
  • Discovering IP

    Asset managers and investors can use the information provided in the platform to access new type of information, which is usually hard to access and calculate. There is strong scientific evidence that analysing intangible assets and intellectual property of a company provides valuable information to form investment decisions. These can be used either stand alone or ideally in combination with existing strategies for the purpose of validating decisions. An alert messaging system provides updates on major IP indicator moves from companies in your watchlist.

    Indicators provided in the platform can be used to form variations of the research portfolio or can be integrated in existing investment strategies. Data can be easily exported as CSV files or directly accessed with an API for further use in 3rd party simulation tools.

    A built in Trade Analyzer allows to add smart-beta to your existing portfolio. Previous trade decisions are being augemented with IP based indicators, and the potential excess return is being calculated.


    M&A advisors can utilise the platform to enhance the process of M&A related short-list generation. Traditionally, when Company X wants to expand in a particular technology area and wants to prepare a list of potential acquisition candidates, questions around how to identify targets or how to rank them arise. A classical job for an M&A advisory firm. With our machine learning based approach, an M&A target list is created by analysing the buyers' technological scope by generating a competence cluster based on its intellectual property. Support vector machines are analysing all patent filings and understand what areas the acquirer owns technologies in. For each area, the machine maps every patent of the acquirer with every patent of the potential targets and provides an overview of its relatedness and closeness. The result is a scatter plot, visualising the top M&A targets, the "M&A shortlist".


    The platform provides real-time access to indicators from IP and intangible asset related research papers, which have been implemented into the system. At the time the book is written, four sets of indicators are available, recreated from the following research papers. These have been identified as being especially significant for the purpose of identifying and measuring IP:

  • Mariagrazia Squicciarini, Helene Dernis, Chiara Criscuolo (OECD), "Measuring Patent Quality"

  • The authors of the OECD paper “Measuring Patent Quality” propose patent quality measures with the aim to facilitate analysis both at the level of the individual patent and at the aggregate patent portfolio level. They are intended to help addressing a number of policy- relevant questions, for example, related to: firms innovation strategies and performance; enterprise dynamics, including the drivers of enterprise creation and of mergers and acquisitions; the determinants of productivity; the financing of innovative enterprises; the output of R&D activities and the returns to R&D investments; the depreciation of R&D; the output of universities and of public research organisations.

  • Lauren Cohen, Karl Diether, Christopher Malloy (Harvard), "Misvaluing Innovation"

  • The authors of the paper “Misvaluing Innovation” demonstrate that a firm’s ability to innovate is predictable, persistent, and relatively simple to compute, and yet the stock market ignores the implications of past successes when valuing future innovation. We show that two firms that invest the exact same in research and development (R&D) can have quite divergent, but predictably divergent, future paths. Our approach is based on the simple premise that while future outcomes associated with R&D investment are uncertain, the past track records of firms may give insight into their potential for future success. We show that a long-short portfolio strategy that takes advantage of the information in past track records earns abnormal returns of roughly 11 percent per year. Importantly, these past track records also predict divergent future real outcomes in patents, patent citations, and new product innovations.

  • David Hirshleifer, Po-Hsuan Hsu, Dongmei Li (University of California, University of Hong Kong), "Innovative efficiency and stock returns"

  • The authors of the paper “Innovative efficiency and stock returns” are focusing on the efficiency of turning R&D expenses into IP and patents. They find that innovative efficiency (IE), patents or citations scaled by research and development expenditures, is a strong positive predictor of future returns after controlling for firm characteristics and risk. The IE- return relation is associated with the loading on a mispricing factor, and the high Sharpe ratio of the Efficient Minus Inefficient (EMI) portfolio suggests that mispricing plays an important role. Further tests based upon attention and uncertainty proxies suggest that limited attention contributes to the effect. The high weight of the EMI portfolio return in the tangency portfolio suggests that IE captures incremental pricing effects relative to well-known factors.

  • Daniel Mattes, Jivka Ovtcharova, Gabriele Zedlmayer, "Using Artificial Intelligence to exploit IP related stock market inefficiencies"

  • In our paper we demonstrate that analysing intangible assets of a company in great detail allows building a model to accurately forecast a company’s ability to convert its intellectual property into stock value in the following quarter. We also demonstrate that a company’s AI-IP impact correlates significantly with its future stock returns. Our findings suggest that it is possible to identify potential winners and losers based on IP related information and that such information can add additional value to every asset management strategy.


    Company details can be retrieved by entering a stock symbol or the company’s website. The firms’ detail sheet consists of the following panels:

    Chart panel

    The orange line-chart displays the monthly stock price for the selected company over the chosen period. With the range scale element, the timeframe for the data displayed can be changed. The grey bars show the CII (Current Impact Index) for each month. The white line-chart displays the IPI (IP to Market Value Indicator) expressing the theoretical market value of the company. Additional information like revenues, EBITDA or R&D expenses can be displayed in the chart, too by selecting them in the legend.

    Indicators & Recommendation Panel

    The panel “Indicators & Recommendations” shows the historical indictors for the company. In this example, on September 3rd 2019, NVIDIA’s stock price closed at 167.51. The IPI on that day was 198.55, which was higher than the stock price (indicated with a green “up” arrow), implicating that the stock is currently undervalued. The quality of the model is 0.96, which is very high on the scale from 0.00 – 1.00 and shows us that the LSTM was able to identify a robust prediction model between CII, R&D expenses and revenues to stock price, providing us with a confidence level of the accuracy of the calculated IPI. The CII on that date is with 4.75 above average (1.00), expressing that NVIDIA’s AI-IP has currently more than 4 times the impact than its sector peers. Both Harvard indicators for January-June and July-December equally show that the company is in the Ability(high) and R&D(high) category. As for IE (Innovation Efficiency), NVIDIA is currently not in the average percentiles (neither high or low).

    Harvard Indicator Panel

    In the Harvard Indicator Panel, we can observe the results of the Harvard Good & Bad R&D metrics. In 2019, NVIDIA has been in the Ability(high) and R&D(high) categories for both periods, January to June and July to December. According to their research paper, the prediction index for January to June is based on the actual values from year-2, and the July to December on results from year-1. It can be seen that NVIDIA has been constantly in the highest Ability quintile 4, and in the higher R&D percentiles 8 and 9 (>7), classifying it as Ability(high) and R&D(high).

    Innovative Efficiency Panel

    The Innovative Efficiency Panel shows the results of the IE calculations, specifically the number of patents granted in that year as well as the calculated value of “R&D Capital” (RDC, as discussed above). The current Patents to RDC ratio is in percentile 1, which is average and doesn’t classify NVIDIA in IE(low) or IE(high).

    Income Statement and Balance Sheet Panels

    Further information, especially providing the development of R&D expenses or the value of intangible assets, is provided in the panels “Income Statements” and “Balance Sheets”.

    Patents Panel

    AI-IP and patents relating to the selected stock can be further examined as shown above through the panel “Patents”.


    The first use case of the platform is the IP research capability. A combined collection of more than 200 million scholarly works, more than 110 million patent records and more than 100 million source code repositories could be searched, which represents almost all the patents, scholarly work and open source codes in the world. You would find segments of this data on a variety of other platforms like Google Patents or Microsoft Academic, but the main differentiator is that all data points have been associated and linked with companies (public and stock listed). This allows the user to, for example, find all AI-IP related patents belonging to a company, a sector or an entire industry.

    Data is being collected from scholarly, patent, source code and financial data providers and includes sources like PubMed, Microsoft Academic, Crossref, USPTO, Orcid, OpenCitations, DOAJ, WorldCat, PMC, UnpayWall, GRID, WIPO, Core, GitHub, Nasdaq, CME, Quandl or Compustat. After logging into the platform, comprehensive searches can be performed.

    It is also possible to review IP related to a specific company. For example, if we want to see AI-IP for the semiconductor company NVIDIA Corporation, we can search for the company by entering its stock symbol NVDA and click on the result. By clicking on a patent, the details are being shown, including the entire patent application, the images, claims and classifications. With a few clicks, access to AI-IP based on individual stock titles and private companies can be discovered.

    Request accesss to the AI Analyst™

    Our platform is available to selected customers. Get in touch to discuss pricing and terms.

    Competence in Artificial Intelligence

    Rigourous systematic.

    Daniel Mattes Daniel Mattes Daniel Mattes