TikTok is a social media platform that provides a real-time window into public sentiment and opinions. Here, users freely express their views on various topics, including brands, products, and financial markets.
What users express via comments or videos on TikTok is sourced by Stockpulse’s crawlers. But how can the right videos be identified? Videos but also comments that matter and thus worth turning into data? On TikTok, videos gain significance only after garnering substantial views. These are the videos we have focused on in the study reported here. We have sourced publicly available comments and videos into separate data sets. Data collection was guided by hashtags and influencers. Firstly, we acquired TikTok videos that fell under hashtags commonly associated with finance, such as #finance, #investment, #money, #crypto, or in the fashion do- main, including #fashion, #style, #shopping, #fashionaddict. In order to complement this dataset, we secondly included videos linked to influencers in finance and fashion, particularly those with a substantial following. This approach led to the identification of thousands of influencer accounts from which we sampled our data.
A key objective of our work at Stockpulse is to offer data that is economically meaningful and actionable. Therefore, we preprocess the data so that it is ticker-mapped and point-in-time with all asset classes being covered. In the case of videos, data processing involved transcribing videos into text and then converting them into data, e.g., on sentiment scores. Our ticker-mapping, key event, and topic-matching methodologies are then applied to extract meaningful datasets.
