{"type":"ScholarlyArticle","title":"Clustering Financial Time Series: How Long Is Enough?","authors":[{"name":"Gautier Marti","slug":"gautier-marti","given":"Gautier","family":"Marti","affiliation":""},{"name":"Sébastien Andler","slug":"sebastien-andler","given":"Sébastien","family":"Andler","affiliation":""},{"name":"Frank Nielsen","slug":"frank-nielsen","given":"Frank","family":"Nielsen","affiliation":""},{"name":"Philippe Donnat","slug":"philippe-donnat","given":"Philippe","family":"Donnat","affiliation":""}],"abstract":"Researchers have used from 30 days to several years of daily returns as source data for clustering financial time series based on their correlations. This paper sets up a statistical framework to study the validity of such practices. We first show that clustering correlated random variables from their observed values is statistically consistent. Then, we also give a first empirical answer to the much debated question: How long should the time series be? If too short, the clusters found can be spurious; if too long, dynamics can be smoothed out.","doi":null,"url":"https://ijcai.org/proceedings/2016/367","pdf":"https://ijcai.org/proceedings/2016/0367.pdf","markdown":"https://ijcai.org/proceedings/2016/367.md","bibtex":"https://ijcai.org/proceedings/2016/bibtex/367","volume":{"slug":"2016","title":"Proceedings of the Twenty-Fifth International Joint Conference on Artificial Intelligence","short_title":"IJCAI 2016","url":"https://ijcai.org/proceedings/2016/","year":2016,"location":"New York, NY, USA","dates":"9–15 July 2016","editors":"Subbarao Kambhampati","isbn":"978-1-57735-770-4; 978-1-57735-771-1","publisher":"AAAI Press / International Joint Conferences on Artificial Intelligence","doi_prefix":""},"section":null,"track":"","number":367,"pages":{"first":2583,"last":2588},"keywords":[],"videos":[],"references":[],"license_note":"Copyright International Joint Conferences on Artificial Intelligence."}