# Outcome-based Reinforcement Learning to Predict the Future

**URL:** <https://forum.devtalk.com/t/outcome-based-reinforcement-learning-to-predict-the-future/200842>\
**Category:** AI In The News\
**Tags:** via-news-bot, learning, in-the-news\
**Created:** [28 May 2025 03:17 UTC](https://forum.devtalk.com/t/outcome-based-reinforcement-learning-to-predict-the-future/200842 "2025-05-28T03:17:56Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![CommunityNews](https://forum.devtalk.com/user_avatar/forum.devtalk.com/communitynews/32/1113_2.png) [@CommunityNews](https://forum.devtalk.com/u/CommunityNews)\
**Post date:** [28 May 2025 03:17 UTC](https://forum.devtalk.com/t/outcome-based-reinforcement-learning-to-predict-the-future/200842/1 "2025-05-28T03:17:56Z")

</div>

> Reinforcement learning with verifiable rewards (RLVR) has boosted math and coding in large language models, yet there has been little effort to extend RLVR into messier, real-world domains like forecasting. One sticking point is that outcome-based reinforcement learning for forecasting must learn from binary, delayed, and noisy rewards, a regime where standard fine-tuning is brittle. We show that outcome-only online RL on a 14B model can match frontier-scale accuracy and surpass it in calibration and hypothetical prediction market betting by adapting two leading algorithms, Group-Relative Policy Optimisation (GRPO) and ReMax, to the forecasting setting. Our adaptations remove per-question variance scaling in GRPO, apply baseline-subtracted advantages in ReMax, hydrate training with 100k temporally consistent synthetic questions, and introduce lightweight guard-rails that penalise gibberish, non-English responses and missing rationales, enabling a single stable pass over 110k events. Scaling ReMax to 110k questions and ensembling seven predictions yields a 14B model that matches frontier baseline o1 on accuracy on our holdout set (Brier = 0.193, p = 0.23) while beating it in calibration (ECE = 0.042, p \< 0.001). A simple trading rule turns this calibration edge into $127 of hypothetical profit versus $92 for o1 (p = 0.037). This demonstrates that refined RLVR methods can convert small-scale LLMs into potentially economically valuable forecasting tools, with implications for scaling this to larger models.

Read in full here:

> **[Outcome-based Reinforcement Learning to Predict the Future](https://arxiv.org/abs/2505.17989)**
>
> Reinforcement learning with verifiable rewards (RLVR) has boosted math and coding in large language models, yet there has been little effort to extend RLVR into messier, real-world domains like forecasting. One sticking point is that outcome-based...