Saved in:
Bibliographic Details
Main Authors: Cui, Yanwei, Wang, Guanghui, Zhang, Xing, He, Peiyang, Li, Ziyuan, Zhu, Bing, Qiu, Wei, Wang, Xusheng, Yu, Zheng, Xin, Anqi
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.23988
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914509604519936
author Cui, Yanwei
Wang, Guanghui
Zhang, Xing
He, Peiyang
Li, Ziyuan
Zhu, Bing
Qiu, Wei
Wang, Xusheng
Yu, Zheng
Xin, Anqi
author_facet Cui, Yanwei
Wang, Guanghui
Zhang, Xing
He, Peiyang
Li, Ziyuan
Zhu, Bing
Qiu, Wei
Wang, Xusheng
Yu, Zheng
Xin, Anqi
contents Time series models predict numbers; decision-makers need advisory -- directional signals with reasoning, actionable suggestions, and risk management. Training language models for such predictive advisory faces a fundamental challenge: quality depends on outcomes unknown at prediction time. We bridge two ideas from reinforcement learning -- using information unavailable during execution to retrospectively generate training signal, and preference alignment -- and propose Hindsight Preference Optimization: observed outcomes let an LLM judge rank candidate advisories on dimensions that scalar metrics cannot capture, producing preference pairs for DPO without human annotation. We apply this to Vision-Language-Model-based predictive advisories on S&P 500 equity time series, demonstrated by a 4B model outperforming its 235B teacher on both accuracy and advisory quality.
format Preprint
id arxiv_https___arxiv_org_abs_2604_23988
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Hindsight Preference Optimization for Financial Time Series Advisory
Cui, Yanwei
Wang, Guanghui
Zhang, Xing
He, Peiyang
Li, Ziyuan
Zhu, Bing
Qiu, Wei
Wang, Xusheng
Yu, Zheng
Xin, Anqi
Machine Learning
Artificial Intelligence
Time series models predict numbers; decision-makers need advisory -- directional signals with reasoning, actionable suggestions, and risk management. Training language models for such predictive advisory faces a fundamental challenge: quality depends on outcomes unknown at prediction time. We bridge two ideas from reinforcement learning -- using information unavailable during execution to retrospectively generate training signal, and preference alignment -- and propose Hindsight Preference Optimization: observed outcomes let an LLM judge rank candidate advisories on dimensions that scalar metrics cannot capture, producing preference pairs for DPO without human annotation. We apply this to Vision-Language-Model-based predictive advisories on S&P 500 equity time series, demonstrated by a 4B model outperforming its 235B teacher on both accuracy and advisory quality.
title Hindsight Preference Optimization for Financial Time Series Advisory
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2604.23988