With the rapid development of large language models (LLMs) in legal applications, systematically evaluating their reasoning ability in judgment prediction has become increasingly urgent. Currently, ...