Large language models (LLMs) often suffer from capability stagnation in self-improvement training due to fixed difficulty levels. To address this issue, we propose STRETCH (Self-Taught Reasoning Evolution via Targeted CHallenge) inspired by cognitive scaffolding theory. STRETCH introduces a dynamic Stretch Zone mechanism that continuously aligns question difficulty with the model’s solving capability. Within a single parameter space, the model alternates between a Scaffolder that generates adaptive challenges and a Learner that solves them through reinforcement learning. This dual-loop co-evolution process stabilize training and promote progressive reasoning growth. Experiments on both negotiation and operation research benchmarks demonstrate that STRETCH consistently outperforms strong prompting and domain-specific baselines. Further analysis shows that dynamic difficulty alignment is critical for sustained capability improvement and synchronized reasoning evolution.