Emotional turmoil is a major risk during crises; it negatively impacts social media communities’ well-being. To reduce this risk, crisis management models have identified emotions specific to each crisis phase. We explore the emotions individuals collectively express and experience on social media by analysing 4,972 videos and 433,097 comments from YouTube surrounding the GME Short Squeeze crisis. We propose a novel methodology involving a RoBERTa model trained on emotions and a KMeans clustering algorithm to identify crisis phases computationally. We found that the community reacted in alignment with the Crisis and Emergency Risk Communication (CERC) model. Emotions of Affection and Happiness were promoted by the community during and post crisis with minimal external influence. We may infer from these findings that social media communities are somewhat self-organising entities promoting good mental health practices and independently encouraging good health and well-being in times of crisis.