# ARC: Augmented-Rank Conformalization for Changepoint Localization --- Finite-Sample Validity and Distribution-Robust Efficiency

arXiv: https://arxiv.org/abs/2608.08424 · HTML: https://mlsanity.com/paper/2608.08424/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 21 of its 42 references that are in the map.
- First posted on arXiv 9 Aug 2026.

## Builds on (most cited first, 21 of 21)

- [An Evaluation of Change Point Detection Algorithms](https://mlsanity.com/paper/2003.06222/) (2020)
- [Testing for a Change in Mean After Changepoint Detection](https://mlsanity.com/paper/1910.04291/) (2019)
- [Multiscale Change-Point Inference](https://mlsanity.com/paper/1301.7212/) (2013)
- [Narrowest-Over-Threshold Detection of Multiple Change-points and Change-point-like Features](https://mlsanity.com/paper/1609.00293/) (2016)
- [Changepoint Detection in the Presence of Outliers](https://mlsanity.com/paper/1609.07363/) (2016)
- [Offline changepoint localization using a matrix of conformal p-values](https://mlsanity.com/paper/2505.00292/) (2025)
- [Online Kernel CUSUM for Change-Point Detection](https://mlsanity.com/paper/2211.15070/) (2022)
- [Post-detection inference for sequential changepoint localization](https://mlsanity.com/paper/2502.06096/) (2025)
- [Theoretical guarantees for change localization using conformal p-values](https://mlsanity.com/paper/2510.08749/) (2025)
- [Robust mean change point testing in high-dimensional data with heavy tails](https://mlsanity.com/paper/2305.18987/) (2023)
- [Distribution-free changepoint localization after sequential change detection](https://mlsanity.com/paper/2606.01256/) (2026)
- [Narrowest Significance Pursuit: inference for multiple change-points in linear models](https://mlsanity.com/paper/2009.05431/) (2020)
- [Online Detection of Changes in Moment-Based Projections: When to Retrain Deep Learners or Update Portfolios?](https://mlsanity.com/paper/2302.07198/) (2023)
- [TUNE: Algorithm-Agnostic Inference after Changepoint Detection](https://mlsanity.com/paper/2409.15676/) (2024)
- [ART: Distribution-Free and Model-Agnostic Changepoint Detection with Finite-Sample Guarantees](https://mlsanity.com/paper/2501.04475/) (2025)
- [Sequential Change Point Detection with FDR Control in Reconfigurable Sensor Networks](https://mlsanity.com/paper/2504.06526/) (2025)
- [Nonparametric Multi Change Point Detection for Markov Chains via Adaptive Clustering](https://mlsanity.com/paper/2607.12369/) (2026)
- [Post Hoc Inference for Component Attribution in Multivariate Change-Point Detection](https://mlsanity.com/paper/2607.14814/) (2026)
- [Beyond Euclidean Summaries: Online Change Point Detection for Distribution-Valued Data](https://mlsanity.com/paper/2602.07252/) (2026)
- [Change-Point Detection for Object-valued Time Series](https://mlsanity.com/paper/2606.00858/) (2026)
- [Change-Aware Self-Adaptive AI-Aided Kalman Filters With Neural Change Point Detection](https://mlsanity.com/paper/2607.13387/) (2026)

## Abstract (arXiv)

Conformal changepoint localization turns any score into a confidence set for the changepoint with finite-sample coverage. Coverage is universal; efficiency is not. The oracle score is a likelihood ratio, so practical scores estimate density ratios, and set length deteriorates under heavy tails, skewness, and distribution shift, where no length guarantee applies. We propose ARC (Augmented-Rank Conformalization), a family of scores depending on the data only through within-segment ranks: rank-CUSUM location and scale channels, their fixed combinations, and a lightweight neural score frozen after synthetic training. Every ARC score inherits finite-sample coverage for every frozen weight configuration, including random initialization and mistraining. The main result is an efficiency transfer theorem: the entire ARC confidence set is almost surely invariant under strictly increasing marginal transforms, so the set length distribution depends on the data pair only through its rank structure, and lengths certified once hold verbatim across its monotone orbit, whereas a plug-in score's length changes with every re-expression. Across different rank structures lengths do change, and are reported as such. Classical rank-test theory positions ARC as targeting the optimal invariant score at bounded cost. Simulations confirm nominal coverage for all scores, including sabotaged networks, identical sets under monotone transforms where plug-in scores inflate, and smooth degradation where plug-in sets become vacuous; on the well-log benchmark ARC localizes annotated shifts to three to five candidates and flags misfit by an empty set. Two boundaries are stated rather than hidden: serial dependence destroys exactness, and trend-type alternatives lie outside the piecewise-exchangeable model.