Paper: arXiv 2310.01285
Authors: Qinmeng Luan, James Hamp
Abstract
Recent work has proposed Wasserstein k-means (Wk-means) clustering as a powerful method to identify regimes in time series data, and one-dimensional asset returns in particular. In this paper, we begin by studying in detail the behaviour of the Wasserstein k-means clustering algorithm applied to synthetic one-dimensional time series data. We study the dynamics of the algorithm and investigate how varying different hyperparameters impacts the performance of the clustering algorithm for different random initialisations. We compute simple metrics that we find are useful in identifying high-quality clusterings. Then, we extend the technique of Wasserstein k-means clustering to multidimensional time series data by approximating the multidimensional Wasserstein distance as a sliced Wasserstein distance, resulting in a method we call `sliced Wasserstein k-means (sWk-means) clustering’. We apply the sWk-means clustering method to the problem of automated regime detection in multidimensional time series data, using synthetic data to demonstrate the validity of the approach. Finally, we show that the sWk-means method is effective in identifying distinct market regimes in real multidimensional financial time series, using publicly available foreign exchange spot rate data as a case study. We conclude with remarks about some limitations of our approach and potential complementary or alternative approaches.
Complexity vs Empirical Score
- Math Complexity: 8.5/10
- Empirical Rigor: 6.0/10 (capped from 7.0: no out-of-sample test; see penalty flags)
- Quadrant: Holy Grail — high math complexity, high empirical rigor (effective empirical rigor capped from 7.0)
Why this score: The paper employs advanced mathematics including Wasserstein distances, barycenters, and sliced approximations, while also demonstrating the method on synthetic and real financial data with hyperparameter tuning and performance metrics.
Research Flowchart
flowchart TD
A["Research Goal:<br>Automated Regime Detection in<br>Multidimensional Time Series"] --> B
subgraph B ["Methodology Development"]
B1["Wk-means Clustering<br>on 1D Synthetic Data"] --> B2{"Performance Analysis"}
B2 --> B3["Identify Metrics for<br>High-Quality Clusterings"]
B3 --> B4["Extend to sWk-means<br>for Multidimensional Data"]
end
B4 --> C["Validation"]
subgraph C ["Data Application"]
C1["Synthetic Multidimensional Data"] --> C2["sWk-means Clustering"]
C3["Real FX Market Data"] --> C2
end
C2 --> D["Key Findings"]
subgraph D ["Outcomes"]
D1["Validated sWk-means on<br>Synthetic Data"]
D2["Identified Distinct<br>Market Regimes in FX Data"]
D3["Established Workflow for<br>Automated Regime Detection"]
end
Evidence Card
What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card
Proposes
The paper introduces sliced Wasserstein k‑means (sWk‑means) clustering for multidimensional time‑series regime detection.
Source sentence
we extend the technique of Wasserstein k-means clustering to multidimensional time series data by approximating the multidimensional Wasserstein distance as a sliced Wasserstein distance, resulting in a method we call `sliced Wasserstein k-means (sWk-means) clustering'.
Assumes 4
- Synthetic returns follow geometric Brownian motion with normally distributed log‑returns. model
Source sentence
Geometric Brownian motion paths S t/S t−1 = exp(rS t ) with parameters Θ are constructed from log returns rt distributed according to rS t ∼N ( (µ −σ2/2)dt, σ2dt ).
- The synthetic data generation uses fixed regime locations and a random offset for the lifting transformation. data
Source sentence
... we use a random offset 0 ≤δ ≤h2 −1 to the lifting transformation for each clustering run.
- The method assumes the Wasserstein barycentre exists and is easily computed for d=1. model
Source sentence
The existence of such a barycentre for the Wasserstein metric, which is easily computed for d = 1, is one of the advantages of using the Wasserstein metric to formulate a k-means clustering algorithm.
- Low‑data environments affect clustering accuracy, implying more data improves performance. statistical
Source sentence
All else being equal, the k-means clustering algorithm is a data‑hungry method, benefitting in terms of performance when supplied with more data, and decreasing h2 allows us to achieve this, which can be particularly beneficial in small data environments.
Data
- Synthetic 1‑d time series synthetic unknown
Source sentence
We consider a period of 20 years, with 252 days in a year and 7 (hourly) observations per day.
- Synthetic 2‑d time series synthetic unknown
Source sentence
We generate synthetic data in a manner similar to that described in Section 3.1.1, with the method extended to two dimensions.
- Foreign exchange spot rate data other public
Source sentence
using publicly available foreign exchange spot rate data as a case study.
Costs
Not applicable
Finds 3
- Median and maximum clustering accuracies increase with larger window size h1 and smaller lifting size h2, especially for reduced‑data sets. statistical
Source sentence
As a broad trend, we observe that the median and maximum accuracies increase…
- The sWk‑means algorithm performs well on synthetic multidimensional data. performance
Source sentence
in Section 3.2.2, we show that the sWk-means algorithm performs well on the synthetic data.
- The sWk‑means method effectively identifies distinct market regimes in real multidimensional financial time series. performance
Source sentence
the sWk-means method is effective in identifying distinct market regimes in real multidimensional financial time series
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 25 related papers). Coverage is partial, so this is neither replication nor refutation.
Code
No repository by the authors found in the paper.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.