Forecast collapse of transformer-based models under squared loss in financial time series

By Pierre Andreoletti

Published 2026-04-02

Everscope rating

1876.3

Relevance to quantitative trading

9 / 10

Implementation complexity

5 / 10

Reproducibility

3 / 5

About this paper

Methodology: Process-level theoretical analysis of ERM under squared trajectory loss with numerical validation. Problem types: Time Series Forecasting, Regression, Sequence-to-Sequence Learning.

arXiv:2604.00064 · Paper rankings

Open the interactive Everscope explorer for full analysis, charts, and paper battles.