parthivnaresh / earningstranscript_nlp

Parsing Earnings Transcripts

The Background

Every quarter, a public company is required to submit a report of its past quarter's financial performance and will usually provide some sort of guidance. Alongside a 10-Q (for U.S.-based companies) submission to the SEC, key management will also hold an earnings call, in which they will verbally describe their performance, future plans, and provide any updates. They will also take questions from analysts who try to further gauge the position of the company. A transcript of this earnings call is available soon after it takes place.

The Significance

The earnings call provides one of the few human interactions with company management that investors, both current and prospective, get to witness. Although the performance of a company in the quarter can affect how the market trades its stock at market open, it's often the future guidance given that determines how the stock moves the next day. These quarterly reports are the most significant information that the public receives about the company and outside of company-specific events or macroeconomic, market-moving news, they are the main pieces of data the market digests as it trades the stock in the ensuing weeks.

The Potential

It's the job of management to stress the good news and downplay the bad. The language that's used (or not used) can be indicative of how management feels about their performance so far and where things are headed. The earnings call is broken into two main sections, prepared remarks and the Q&A section. The prepared remarks are chosen before hand and are more structured. The Q&A section consists of spontaneous responses to analyst questions. From an NLP perspective, each section has its advantages and disadvantages. While more indicate of a company's performance, the prepared remarks section of a dismal quarter can seem great due to the flowery language used by management. The Q&A section contains more spontaneous and genuine commentary but there can be greater variance since some executives are more experienced at hiding the truth than others. Using both sections and finding ways to mitigate their downsides seems like the best way to get the complete picture.