Artificial intelligence makes portfolio management more efficient and enables creating new investment products. Portfolio manager Mattias Lagerspetz shares how to leverage AI in investing – and how, for example, he used it to build a customized vegan portfolio for a client.
Portfolio manager Mattias Lagerspetz spends his workdays on rather unexpected tasks.
“Mostly, I code”, he says.
Lagerspetz is part of a four-person team developing AI-assisted portfolio management at Evli. The team manages just over one billion euros in assets with the support from Atlas Intelligence, an in-house AI tool.
Lagerspetz leverages AI in something traditionally done by humans: qualitative company analyses that are the foundation of investment decisions.
There has been a significant jump in AI usage over the past decade. Before, AI was mainly used for quantitative analyses where decisions are based on data and statistical models. Now, AI can also incorporate qualitative factors, such as sustainability.
“The simpler and more repetitive the task, the more value there is to automation.”
Lagerspetz doesn’t believe that AI is better at investing than humans. However, it can replicate human-designed processes thousands of times, tirelessly and without complaints.
“The simpler and more repetitive the task, the more value there is to automation. Automation doesn’t make sense if a task is done seldom or is one-off.”
The most challenging tasks to automate are those that require human interaction or an in-depth understanding of concepts like organisational practices.
“In the early stages of the investment process, brainstorming usually requires lengthy discussion both internally and with clients to gauge their needs.”
Data, processes, and experience enable the use of AI
According to Lagerspetz, using AI efficiently in portfolio management requires three things: controlled and high-quality data, a clear and structured process and the portfolio manager’s expertise in complex cases.
The portfolio manager starts by defining the workflow and relevant, company-specific input data. Getting data from the right sources is crucial.
“In comparison to using ChatGPT in daily life, I want much more control over the AI. The data I use is based on high-quality sources defined by us, not the AI’s possibly erroneous data.”
Another success factor is prompt engineering, or how questions are asked. If the questions are too vague, the answer is based on the AI’s intuition and the internet’s collective opinion.
“We want the company analysis to reflect our vision. We need to break the questions into as small pieces as possible.”
For example, Lagerspetz and his team have tailored a client portfolio focused on defence-related issues and preparation for U.S. tariffs. Instead of asking the AI whether a company is vulnerable to trade wars, which would yield a vague answer, they asked: “From which regions does the company generate revenue?” and “Is the production or imports dependent on the U.S. or China?”
“This way, we got structured, reasoned and comparable answers,” Lagerspetz explains.
The final touch to AI’s input is the portfolio manager’s expertise. If AI’s answer points to several directions, Lagerspetz steers it based on his experiences of similar situations.
“In other words, I guide the AI just like I would a newly hired analyst at Evli.”
Using AI to tailor client portfolios
AI has enabled Evli to create an entirely new product: customised investment portfolios. It means tailoring a portfolio to a client’s preferences, values and views.
“For example, we had a client who wanted a vegan investment portfolio,” Lagerspetz explains.
This request was so rare that there were no pre-existing datasets to base decisions on. The AI had to be guided carefully.
“If I ask ChatGPT whether a company is vegan, it can’t say. There are many angles to consider: Is an airline vegan if it offers vegan food? Can a car manufacturer be vegan? What if the seats are made of leather?”
AI is currently only used in a small part of the team’s investments.
“If there is no need to weigh the portfolio thematically, we may only use AI in specific areas.”
While the customised portfolios are now only offered to large investors, Lagerspetz believes AI could be beneficial for smaller clients as well.
The AI tool, developed by the team, is also used outside of Evli. The latest partnership is with the Global Child Forum, an international foundation founded by the Swedish royal family. The foundation promotes children's rights in the corporate world.
Evli has offered the tool for the use of the Global Child Forum. With it, the foundation can analyse how roughly 2,000 global companies address children's rights in their operations more efficiently and in more depth.
“The goal is to help the Global Child Forum produce more accurate information for companies and investors. At the same time, the employees are freed up to develop new things and expand the foundation.”
The future of AI
New technology is addictive. At least Lagerspetz admits he is hooked.
“I get to develop new things and be at the forefront, especially regarding new technologies.”
Lagerspetz is hopeful about the future of AI. He compares it to a new employee who can free the portfolio manager from repetitive data handling and other manual tasks.
“Many believe that there is nothing AI cannot do in the future. I don’t entirely share this sentiment. The world is a pretty complex interface, after all.”
4 tips for using AI in investing
1: Brainstorm ideas
AI can be useful in getting started and generating ideas. You can ask, for example, how a company would fare when tariffs between the U.S. and China come into effect.
2: Break down your questions
Pay attention to how you word your prompts. Avoid too broad questions, like “Which companies should I invest in right now?” Be specific and ask questions like “Which companies benefit from the development of AI?”
3: Understand the limitations
Acknowledge AI’s limits in investing. While progress is being made, not all AI-based investment tools work well at scale. Current AI-based company analysis services may still require professional oversight and support from the service provider. This means they are not yet fully automated plug-and-play solutions.
4: Ask for help from a professional
If you don’t have the time, knowledge or interest to make informed investment decisions, consulting an asset manager is a smart move. A professional supports you in building a diversified portfolio, following the markets, making decisions and considering individual factors, such as goals, taxation and estate planning.
Tips provided by Evli’s portfolio manager Mattias Lagerspetz.
