In the dynamic landscape of cybersecu-rity and cyber warfares, Cyber Threat Intelligence (CTI) is increasingly relied on for gathering and sharing the latest information about threats and their trends. Current CTI sharing methods (e.g., ISACs, automated STIX/TAXII platforms), face challenges in terms of scalability, trust, and data quality issues. This is because they often lack systematic metrics for evaluating the quality and relevance of the threat data that are being shared. Moreover, they do not offer any mechanism to enable participating organizations to autonomously make decisions as to what Threat Intelligence providers to request and share data from/to. To address these limitations, we propose a novel Threat Intelligence sharing approach based on coalitional game theory. We first propose a set of metrics that enable organizations to assess the effectiveness of the shared Threat Intelligence data. Based on these metrics, we propose a preference function and a coalition formation algorithm that enable organizations to autonomously join and leave Threat Intelligence coalitions until reaching a Nash-Stable situation wherein no organization has incentive to leave its current coalition and join another one. Experiments suggest that our solution significantly improves the Mean Time to Detect (MTTD), Mean Time to Respond (MTTR) and Containment Rate.