Earnings call is a presentation of the company that encompass not only the financial results but also the detailed explanation that may not appear on numbers. This paper investigates earnings call transcripts of listed companies and look for non-quantified signals that could explain the market better in terms of market trends and sector grouping with the use of topic modeling. With the use of topic modeling, we generate topics, which are grouped tokens that co-occurs the most frequently, that could possibly represent the market for each quarter. After evaluating the generated topics, this paper finds that the change in occurrence of tokens within the topic represents some market trend. Furthermore, the stocks, when grouped by the topic, showed comparable performance to the industry sector groups. We thus conclude that the topic modeling approach can identify implicit information that are not shown in the financial data.