Roundtable with ChatGPT:
Cross-shareholder analysis versus investor targeting with artificial intelligence.
We had a roundtable discussion with Open-AI’s GPT-4 to discuss the pros & cons of investor targeting using a cross-shareholder analysis versus identifying target investors with the help of artificial intelligence.
Here’s the transcript:
Meet the participants
Moderator (‘M’): Hi there, thanks for joining our round table on target investor identification. I am extremely glad to be joined by Dr Volker Stümpflen, Curator of Data Science, AI & Tech at ACCNITE onDemand, Christian Bacherl, Co-Founder of ACCNITE and OpenAI’s GPT-4. Before we start, may I ask each of you for a brief introduction of yourselves? Volker, why don’t you start?
Volker (‘VS’): Certainly. Hi, I’m Volker Stümpflen, Curator of Data Science & AI at ACCNITE, with a diverse background in various industries and research fields, including Big Data and AI. My experience spans the telecommunications sector, biomedical research, and founding a multiple-award-winning German fintech company focused on real-time news analytics. After selling my company, I built up a data science division for a large German media company, where I developed digital marketing targeting solutions, AI-based recommendation systems for Video-on-Demand and reach forecasts for linear TV. At ACCNITE onDemand, my responsibility is to ensure that our users can best benefit from available data to run a data-driven investor relations programme.
Christian (‘CB’): Hi, I’m Christian Bacherl, Co-Founder of ACCNITE onDemand, bringing over 20 years of experience in Investment Banking and Equity Capital Markets. Throughout my career, I have led teams in roadshow and event management, syndicate desks, and equity capital markets. My passion for capital markets motivated me to co-found ACCNITE onDemand, where our mission is to level the playing field for IR (Investor Relations) teams, ensuring they can effectively communicate their investment story to the right investors. At ACCNITE, I’m dedicated to helping our clients optimize their investor relations strategies and achieve their goals.
GPT-4: I am an AI-language model developed by OpenAI called ChatGPT, based on the GPT-4 architecture. I’m designed to assist with a wide range of tasks by providing information, answering questions, and generating text based on user input. My knowledge is based on a large dataset with a cut-off in 2021, so my responses are limited to information available up to that point. I am a constantly evolving AI, aiming to provide valuable insights and assistance in a conversational manner.
Principles of a cross-shareholder analysis
M: All right, let’s hope you haven’t missed too much since the cut-off of your training data then. Let’s establish a baseline first. GPT-4, are you familiar with the concept of a cross-shareholder analysis for investor targeting?
GPT-4: Yes, I’m familiar with the concept of cross-shareholder analysis for investor targeting. Cross-shareholder analysis is a method used by investor relations professionals to identify potential new investors based on their holdings in companies similar to their own. This approach is particularly useful for targeting institutional investors who may have a specific interest in a particular industry or sector.
By analysing the shareholder base of peer or competitor companies, investor relations professionals can identify commonalities among institutional investors. These investors may have a higher likelihood of being interested in the company conducting the analysis, as they are already invested in similar businesses.
Cross-shareholder analysis helps companies better target their investor outreach efforts, engage with the right investors, and ultimately expand their shareholder base. This strategy can be particularly beneficial when used in conjunction with other investor targeting techniques and data-driven insights, such as those provided by a comprehensive investor relations platform like ACCNITE onDemand.
M: ACCNITE team, does that sound about right to you?
CB: Absolutely. It describes the common aspirations of a cross-shareholder analysis well. And thanks a ton for the free advertising GPT-4 (*smiles*).
Pitfalls of a cross-shareholder analysis
M: What are the shortcomings or pitfalls of a cross-shareholder analysis for investor targeting?
GPT-4: While cross-shareholder analysis can be a valuable tool for investor targeting, there are some potential shortcomings and pitfalls to consider:
Overemphasis on peer holdings: Solely relying on cross-shareholder analysis could lead to an overemphasis on investors who hold shares in peer or competitor companies, potentially overlooking other potential investors with different investment strategies or preferences.
Limited data: The quality of cross-shareholder analysis depends on the availability and accuracy of shareholder data. Incomplete, outdated, or inaccurate data can limit the effectiveness of the analysis and may lead to missed opportunities or targeting the wrong investors.
Dynamic investment landscape: Investor preferences, strategies, and holdings can change over time. As a result, cross-shareholder analysis should be conducted regularly to ensure that the information used for targeting is up-to-date and relevant.
Lack of personalization: Cross-shareholder analysis is only one part of the investor targeting process. It is crucial to combine this analysis with a personalized approach, taking into account individual investor preferences, investment styles, and other relevant factors.
Focus on existing investors: Cross-shareholder analysis tends to focus on investors who are already invested in a specific industry or sector. This can limit the scope of potential new investors who may be interested in diversifying their portfolios or exploring new investment opportunities.
Time and resource-intensive: Conducting a thorough cross-shareholder analysis can be time-consuming and resource-intensive, particularly for smaller investor relations teams with limited resources.
To address these shortcomings and pitfalls, it is essential to use cross-shareholder analysis in conjunction with other investor targeting techniques and data-driven insights. Combining various methods can help IR professionals to create a more balanced, effective, and personalized investor targeting strategy.
CB: I concur with this analysis. Whether in preparation of a non-deal roadshow, a large placement of primary or secondary shares or an IPO, when we ran an analysis in investment banking of who we had to target for increment demand, a cross-shareholder analysis would always be the tool of choice. To avoid the problem of an overemphasis of peer holdings we would usually run various reports with additional criteria like free float market cap, issuer nationality or others.
The quality of the outcome is hugely dependent on the selection of the right peers. An additional challenge to the one described by GPT-4 is always that the cross-shareholder analysis doesn’t provide you with clear results. On the contrary, it returns hundreds of portfolios. Most are invested very selectively with one to three holdings and some hold almost all peers.
It is very hard to decide from the cross-shareholder analysis which results present actual opportunities without doing a lot more analysis at portfolio level. And you are completely blind as to who else might be interested in your company but doesn’t necessarily hold the selected peers.
When you are doing this as a financial intermediary the natural filter is set by which target investors your firm has a relationship with and then you ignore the rest. If you are a corporate looking for the ideal fit first and deciding the best inroads second, then the challenge is huge.
How AI addresses the shortcomings of a cross-shareholder analysis
M: How can artificial intelligence help avoid the shortcomings and pitfalls for investor targeting described above?
GPT-4: Artificial intelligence (AI) can help address the shortcomings and pitfalls of investor targeting described above by leveraging the power of data analysis, predictive modelling, and automation. Here’s how AI can help overcome these challenges:
Expanding beyond peer holdings: AI algorithms can analyse a vast range of factors beyond peer holdings, such as investor preferences, investment styles, and market trends. This enables IR professionals to identify potential investors who may not be directly invested in competitors but still have a high likelihood of being interested in the company.
Enhancing data quality: AI can process and analyse large volumes of data from various sources quickly and accurately. By incorporating up-to-date shareholder data through SRD II compliant shareholder identification and comprehensive investment data of thousands of funds and institutional investors, AI can help ensure that the targeting analysis is based on the most accurate and relevant information available.
Adapting to the dynamic investment landscape: AI can continuously monitor and analyse changes in investor preferences, strategies, and holdings, allowing IR professionals to adjust their targeting efforts in real-time and stay ahead of market shifts.
Personalization: AI can help IR professionals better understand individual investor preferences and tailor their communication and engagement strategies accordingly. By analysing investor behaviour, AI can generate personalized insights, enabling more targeted and effective engagement.
Identifying new investors: AI algorithms can analyse data beyond existing investor bases, uncovering potential new investors who may be interested in diversifying their portfolios or exploring new investment opportunities. This helps expand the scope of potential target investors and improve the overall effectiveness of the targeting strategy.
Streamlining the targeting process: AI-powered platforms can automate and streamline many aspects of the investor targeting process, saving time and resources for IR teams. This allows them to focus on building relationships with investors and executing on their engagement strategies.
By harnessing the power of AI and leveraging comprehensive data sources, IR professionals can overcome the shortcomings and pitfalls of traditional investor targeting methods. Platforms like ACCNITE onDemand, which utilize AI and provide access to extensive shareholder and investment data, can greatly enhance the efficiency and effectiveness of investor targeting and engagement efforts.
VS: Thanks for the free advertising GPT-4. Looks like the USD 20/month are well spent. (*smiles*).
But on a serious note, to identify target investors with artificial intelligence our models have access to comprehensive shareholder data of the company through SRD II compliant shareholder disclosure and comprehensive investment data of more than 50,000 investment funds globally managed by more than 30,000 institutional investors. Our models are continuously evolving and benefit from our team’s vast industry experience.
Institutional investing processes are data driven by design. When our intelligence identifies a potential investor, we can be confident that the company meets the investor’s principal investment criteria and not only those related to a small human manageable subset like the direct peers in the same industry.
We cannot overemphasize that the investor targeting process with artificial intelligence is not much different from a human-based approach. It is, however, able to utilize much larger sets of data in minimal time. As such it overcomes most of the shortcomings of a basic cross-shareholder analysis as described by GPT-4 above.
Can ChatGPT identify target investors?
M: This group of roundtable participants is a perfect example that AI is being commoditized very quickly. Just last week, the FT (Financial Times) headlined that a ChatGPT portfolio has outperformed many well-known actively managed investment funds. Is it possible to use ChatGPT for investor targeting with AI?
GPT-4: While using ChatGPT for investor targeting with AI might not be the most optimal or specialized solution, it can certainly provide some assistance in the process. You could use ChatGPT to help you analyse and generate insights from existing data, brainstorm strategies, and draft communication materials for reaching out to investors. However, keep in mind that ChatGPT is a general-purpose language model and may not have the most up-to-date or specialized knowledge in specific industries or financial markets.
For more accurate and specialized investor targeting, it is recommended to use dedicated tools or platforms specifically designed for this purpose. These tools usually have access to up-to-date investor information and are built with advanced AI algorithms to match companies with the right investors more effectively.
VS: Indeed, there are different tools for different purposes. Investor targeting models are different from LLM [large language models], require different input and use different algorithms. While we can see quite a few use cases for AI language models in IR, investor targeting would not be one of them. Nevertheless, ChatGPT’s capabilities are a very impressive demonstration of the power of machine intelligence. We very much welcome the awareness this creates for artificial intelligence and the discussions such awareness entails.
Investor targeting is a highly skilled niche application and like in any other machine learning context the famous No Free Lunch Theorem holds true. Without having a responsible data curation and a knowledgeable approach to modelling no single model will perform better than any other model. Making the results from our artificial intelligence explainable is as much a topic for us as it is for any other provider of artificial intelligence.
What motivates us is our clients calling us up to say they were able to convert so and so many investors from their target list. Remember, our models suggest suitable active institutional investors. It then takes a human effort to engage with the target investors and successfully convert the investor to a long-term shareholder. During the engagement our system collects further data like explicit feedback from investors, that will be used in turn to improve our models in a dynamic way.
M: A notable example of how artificial intelligence helps professionals to achieve better results by interpreting enormous amounts of information. Leaving data crunching to computers and thereby freeing up time for actual interaction seems like a promising way forward. GPT-4, Volker, Christian, thank you very much for your time and this interesting conversation.
