Creating investment portfolios with generative AI
“We will challenge the financial market with the artificial intelligence (AI) capabilities we have built in the advertising market.”
Park Sung-hyuk, CEO of Impact AI, who recently met with Maeil Economy in Silicon Valley, U.S., said, “As a company’s advertising cost execution structure is similar to the automated portfolio investment in the financial market, I think the optimal advertising cost execution model can be applied to investment to achieve results.”
Impact AI is an AI start-up established in 2022 by CEO Park, a professor at KAIST Business School, with co-founder Lee Min-hyung. Based on future demand forecasting AI, it provides a service that tells you how much to spend on each channel to advertise the product most efficiently. CEO Park said, “Different domestic companies are using our AI to optimize advertising budgets executed between multiple platforms.” The financial service that CEO Park is envisioning is to analyze reports prepared by analysts at securities firms using generative AI and create an investment portfolio accordingly. At first glance, it seems clichéd, but in reality, it is not. “The biggest ability of generative AI is that it can read text, grasp the core, and think based on it,” he said. “It is different from existing mathematical models and financial data-oriented AI trading.”
“In the past, machine learning-based trading was to trade with quantitative data released according to a set mathematical investment model,” CEO Park said. “The model we are making is for AI to read hundreds of reports written by analysts, from individual stocks to macroeconomic analysis, find insights, and find positive alpha returns that are not reflected in the market.”
CEO Park said, “We are analyzing securities company analyst reports in the Chinese stock market using generative AI and making actual investments by making a portfolio accordingly.” “So far, AI’s portfolio has performed more than 12% compared to the market.”
[Silicon Valley correspondent Lee Deok-ju]
[Source] https://www.mk.co.kr/en/it/11138810
