An arXiv abstract introduces Paloma, a time series forecasting method built on the idea that explicitly modeling periodicity is efficient. In this setup, the periodic component is removed from the series, and the remaining signal is the residual.
The abstract notes that this residual is typically treated as unstructured. Paloma is described as phase-conditioned residual modulation; the name suggests that the method modulates the residual based on phase information, but the provided text is truncated before further explanation.
The source is an announcement abstract only, with no experimental results or comparisons included. The claims are therefore limited to the method's motivation and naming.