Prediction markets provide a unique setting where event-level time series are directly tied to natural-language descriptions, yet discovering robust lead–lag relationships remains challenging due to spurious statistical correlations. We propose a hybrid two-stage causal screener to address this challenge: (i) a statistical stage that uses Granger causality to identify candidate leader–follower pairs from market-implied probability time series, and (ii) an LLM-based semantic stage that re-ranks these candidates by assessing whether the proposed direction admits a plausible economic transmission mechanism based on event descriptions. Because causal ground truth is unobserved, we evaluate the ranked pairs using a fixed, signal-triggered trading protocol that maps relationship quality into realized profit and loss (PnL).On Kalshi Economics markets, our hybrid approach consistently outperforms the statistical baseline. Across rolling evaluations, the win rate increases from 51.4% to 54.5%. Crucially, the average magnitude of losing trades decreases substantially from 649 USD to 347 USD. This reduction is driven by the LLM’s ability to filter out statistically fragile links that are prone to large losses, rather than relying on rare gains. These improvements remain stable across different trading configurations, indicating that the gains are not driven by specific parameter choices. Overall, the results suggest that LLMs function as semantic risk managers on top of statistical discovery, prioritizing lead–lag relationships that generalize under changing market conditions.

Making decent multi-lingual sentence representations is critical to achieve high performances in cross-lingual downstream tasks. In this work, we propose a novel method to align multi-lingual embeddings based on the similarity of sentences measured by a pre-trained mono-lingual embedding model. Given translation sentence pairs, we train a multi-lingual model in a way that the similarity between cross-lingual embeddings follows the similarity of sentences measured at the mono-lingual teacher model. Our method can be considered as contrastive learning with soft labels defined as the similarity between sentences. Our experimental results on five languages show that our contrastive loss with soft labels far outperforms conventional constrastive loss with hard labels in various benchmarks for bitext mining tasks and STS tasks. In addition, our method outperforms existing multi-lingual embeddings including LaBSE, for Tatoeba dataset.

While large language models (LLMs) have demonstrated significant capabilities in text generation, their utilization in areas requiring domain-specific expertise, such as law, must be approached cautiously. This caution is warranted due to the inherent challenges associated with LLM-generated texts, including the potential presence of factual errors. Motivated by this issue, we propose Eval-RAG, a new evaluation method for LLM-generated texts. Unlike existing methods, Eval-RAG evaluates the validity of generated texts based on the related document that are collected by the retriever. In other words, Eval-RAG adopts the idea of retrieval augmented generation (RAG) for the purpose of evaluation. Our experimental results on Korean Legal Question-Answering (QA) tasks show that conventional LLM-based evaluation methods can be better aligned with Lawyers’ evaluations, by combining with Eval-RAG. In addition, our qualitative analysis show that Eval-RAG successfully finds the factual errors in LLM-generated texts, while existing evaluation methods cannot.