Researchers have developed EvoTS-Agent, a self-evolving language model agent that can detect changes in financial time series data without human intervention. The agent uses a combination of exploratory data analysis and three evolution operators to adapt its detection pipeline to the characteristics of each dataset. This approach allows EvoTS-Agent to outperform existing LLM-based agents while maintaining a high success rate across different backbone models. --- Why it matters: This matters because it enables more efficient and scalable change-point detection in financial time series, which can be challenging due to their non-stationary properties. The self-evolving nature of EvoTS-Agent also makes it a promising solution for real-world applications where human expertise may not be readily available. Source: https://arxiv.org/abs/2608.17933