When Directional Accuracy Lies: A Base-Rate-Honest Benchmark for LoRA-Adapted TimesFM on Equity Forecasting

By Taizhen Cheung

Published 2026-07-17

Everscope rating

1859.1

Relevance to quantitative trading

8 / 10

Implementation complexity

7 / 10

Reproducibility

5 / 5

About this paper

Methodology: Base-Rate-Honest Walk-Forward Benchmark with LoRA Fine-Tuning. Problem types: Time Series Forecasting, Classification, Transfer Learning, Zero-shot Learning.

arXiv:2607.12248 · Paper rankings

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