ML Anthology

Clustering Financial Time Series: How Long Is Enough?

Gautier Marti, Sébastien Andler, Frank Nielsen, Philippe Donnat
IJCAI 2016 pp. 2583-2589
/ijcai/2016/marti2016ijcai-clustering/

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. PDF

PDF IJCAI Semantic Scholar

Cite

Text

Marti et al. "Clustering Financial Time Series: How Long Is Enough?." International Joint Conference on Artificial Intelligence, 2016.

Markdown

[Marti et al. "Clustering Financial Time Series: How Long Is Enough?." International Joint Conference on Artificial Intelligence, 2016.](https://mlanthology.org/ijcai/2016/marti2016ijcai-clustering/)

BibTeX

@inproceedings{marti2016ijcai-clustering,
  title     = {{Clustering Financial Time Series: How Long Is Enough?}},
  author    = {Marti, Gautier and Andler, Sébastien and Nielsen, Frank and Donnat, Philippe},
  booktitle = {International Joint Conference on Artificial Intelligence},
  year      = {2016},
  pages     = {2583-2589},
  url       = {https://mlanthology.org/ijcai/2016/marti2016ijcai-clustering/}
}