arXiv cs.AI· Jaehoon Lee, Jun Seo, Seunghan Lee, Tae Yoon Lim, Dongwan Kang, Hwanil Choi, Minjae Kim, Sungdong Yoo, Junhyeok Kang, Sangjun Han, Soonyoung Lee, Hao Ni, Srijan Sood, Stefan Zohren, Chanyeol Choi, Yongjae Lee, Zhangyang Wang, Lukasz Szpruch, Benjamin Bowring, Wonbin Ahn·· 12 小时前AI 评分38
FinVerse：面向金融时间序列预测的更真实评测基准
FinVerse: Toward More Realistic Evaluation for Financial Time-Series Forecasting
AI 导读
FinVerse 是一个金融时间序列基准，收录 116,897 条序列、171.1M 条观测，按经济语义组织数据，并从逐点预测、横截面排名和组合回测三个维度评估目标序列。
正文 · 原文
Authors:Jaehoon Lee, Jun Seo, Seunghan Lee, Tae Yoon Lim, Dongwan Kang, Hwanil Choi, Minjae Kim, Sungdong Yoo, Junhyeok Kang, Sangjun Han, Soonyoung Lee, Hao Ni, Srijan Sood, Stefan Zohren, Chanyeol Choi, Yongjae Lee, Zhangyang Wang, Lukasz Szpruch, Benjamin Bowring, Wonbin Ahn
Abstract:Time-series foundation models (TSFMs) are increasingly evaluated on broad benchmarks that apply uniform conventional forecast evaluation metrics across heterogeneous time-series. However, in the financial domain, decision-relevant evaluation requires going beyond conventional metrics to assess forecasts in terms of directional accuracy, cross-sectional ranking, and portfolio outcomes. To this end, we introduce FinVerse, a financial time-series benchmark, collecting a broad collection of 116,897 series with 171.1M observations. FinVerse organizes financial data according to their economic semantics and evaluates target series across point-wise forecasting, cross-sectional ranking, and portfolio backtesting. Across 43 public TSFMs, model rankings on FinVerse show only moderate agreement with those on the existing benchmark, GIFT-Eval (r=0.43), suggesting that generic forecasting benchmarks do not fully capture the aspects of forecast quality that matter for financial decision-making. Furthermore, substantial variation in model performance across financial time-series categories and evaluation metrics provides a more informative basis for model selection, enabling practitioners to identify suitable models according to their target series and downstream objectives.
| Comments: | 40 pages |
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2608.03259 [cs.LG] |
| (or arXiv:2608.03259v3 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2608.03259 arXiv-issued DOI via DataCite |
Submission history
From: Jaehoon Lee [view email]
[v1]
Tue, 4 Aug 2026 07:32:42 UTC (106 KB)
[v2]
Thu, 20 Aug 2026 09:12:23 UTC (106 KB)
[v3]
Fri, 23 Oct 2026 03:59:58 UTC (406 KB)
来源：arXiv cs.AI · arxiv.org