Use case: AI-enhanced thematic portfolio construction (ETF selection)
Context:
A leading Swiss private bank needed to automate and enhance the selection of ETFs aligned with specific investment themes, previous attempts failed due to biased data (greenwashing, languages, …).
Objective:
Compute purity scores between an ETF (and its constituents) and a defined theme based on news, patents, and web information.
Technologies:
Python
Natural Language Processing
Web scrapping and data mining
Solution:
A front-end interface enables the selected themes and ETFs, computing the purity scores as well as enabling the deep-dive and interpret the results. Our solutions analyze millions of information and address bias in multiple ways to ensure reliable and accurate insights.