# Optimal Value Inference for Reinforcement Learning

arXiv: https://arxiv.org/abs/2609.09981 · HTML: https://mlsanity.com/paper/2609.09981/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in RL theory and sample complexity, one of 15 directions in Deep reinforcement learning methods.
- Builds on 5 of its 64 references that are in the map.
- First posted on arXiv 9 Sep 2026.

## Placed in

- Deep reinforcement learning methods / [RL theory and sample complexity](https://mlsanity.com/deeprl/rl-theory-and-sample-complexity/)

## Builds on (most cited first, 5 of 5)

- [Reinforcement Learning for Non-Stationary Markov Decision Processes: The Blessing of (More) Optimism](https://mlsanity.com/paper/2006.14389/) (2020)
- [Finite Sample Analysis of Minimax Offline Reinforcement Learning: Completeness, Fast Rates and First-Order Efficiency](https://mlsanity.com/paper/2102.02981/) (2021)
- [Model-Free Non-Stationary RL: Near-Optimal Regret and Applications in Multi-Agent RL and Inventory Control](https://mlsanity.com/paper/2010.03161/) (2020)
- [Inference on Optimal Policy Values and Other Irregular Functionals via Softmax Smoothing](https://mlsanity.com/paper/2507.11780/) (2025)
- [Semiparametric Off-Policy Inference for Optimal Policy Values under Possible Non-Uniqueness](https://mlsanity.com/paper/2505.13809/) (2025)

## Abstract (arXiv)

We study offline inference for the optimal value in reinforcement learning under finite state and action spaces. Two new nuisances are derived as fixed points of a self-induced Bellman equation, in which we approximate the maximum Bellman operator by its softmax correspondence. We propose a debiased estimator through the Neyman orthogonality and establish its asymptotic normality under diverging horizons even when the behavior policy changes with time, as long as the nuisances have the statistical rates that can be achieved by many machine learning methods. We provide a concrete estimating procedure for these nuisances and show they can lead to valid inference. Synthetic experiments validate the numerical performance of our inference method, and we implement it in real-life decision-making problems, including bike repositioning and AI agentic tool use.