Relative Value Learning

[system]

---

This is a companion discussion topic for the original entry at https://arxiv.org/abs/2607.21120