Post-training improves stock price forecasting with language models
Can Language Models Learn to Forecast Stock Prices
Computation and Language
Summary
Predicting stock prices is very tricky because the market is noisy and it's unclear what information matters most. The authors show that a special training process called post-training helps a language model get better at predicting future stock returns by learning to gather useful market information and decide on predictions more effectively. After this training, the model’s accuracy in predicting the direction and size of stock price movements improved significantly. This suggests language models can be adapted to forecast financial markets better than before.
What this means in practice
- •For quantitative trading teams: Use fine-tuned language models to enhance decision-making by improving predictions of stock price direction and magnitude from market data.
- •For financial data engineers: Implement updated language models that better select and query relevant market signals for automated financial forecasting tools.
Authors
Jiacheng Guo, Suozhi Huang, Shuzhen Li, Yunlong Gao, Zerui Cheng, Jason Ge, Shushu Liang, Zihao Li, Hao Lu, Ming Yin, Shilong Liu, Jiashuo Liu, Xu Kuang, Mengdi Wang
Abstract
Post-training has been shown to significantly improve language models' performance on tasks with verifiable outcomes, including mathematical reasoning, software engineering, and computer use. However, whether the same approach can improve forecasting in financial markets is much less clear. Compared with tasks with verifiable outcomes, not only are realized returns noisy, but even what constitutes a relevant information set for making effective predictions is not obvious a priori: the model must decide which observations to gather and then commit to a numerical judgment before the outcome is known. We study this question in a chronological stock-price sandbox, where a language model gathers price, volume, relative-performance, and market-context evidence and predicts a future return. We post-train Qwen3-4B with supervised fine-tuning (SFT) on tool-use demonstrations, then proximal policy optimization (PPO) with a terminal reward given by the forecast score against the realized return. The resulting AURA-4B more than doubles the starting direction--magnitude score, from 20.94 to 43.31, and is comparable to frontier language models on this benchmark. Conditional magnitude agreement rises from 33.3 to 66.2, while directional accuracy changes from 62.9 to 65.4. SFT expands tool use, and PPO further increases the share of ranking and market-context queries. These results show that post-training can substantially improve financial forecasting performance, together with changes in how the model investigates the market, on this outcome-selected benchmark.