5 min read
How equity analysts use sentiment for earnings research
June 10, 2019
4 min read
It’s more important than ever to stay ahead of market-moving trends. The sentiment is one indicator of how management teams are communicating the business impact of major events. By analyzing tonality in millions of earnings call and event transcripts for the past 10+ years, we’ve developed a sentiment scoring model. Try Sentiment within AlphaSense.
How AlphaSense Sentiment work
Sentiment analysis is a type of AI that helps identify, quantify, and analyze levels of emotion within a portion of text. When AlphaSense Sentiment is activated, phrases within Earnings transcript documents are highlighted in green (positive) or red (negative), to indicate statements as either positive or negative.
Above: Spotify Q1 2020 Earnings Call layered with AlphaSense Sentiment Analysis.
Sentence-level sentiment analysis
AlphaSense Sentiment is also capable of selecting which portions of the text are more positive or negative than others, allowing you to quickly get to the core of the statement.
Below, you can also see how Sentiment understands when management is talking about a negative trend in expenses, and can measure the varying degree of negativity within the document. This means you can quickly scan the document and focus on highlighted areas that may be of greater importance to you as an investor or analyst. This can save you valuable time that would be otherwise spent on going through the whole transcript and searching for the same areas.
Above: Ford Q1 2020 Earnings Call layered with AlphaSense Sentiment Analysis.
Understanding Sentiment scores and trends
Sentiment can also act as a check of your previous work. Say you have already gone through the transcript. With Sentiment, you can easily double-check whether you’ve missed any important statements that could impact your investment thesis.
AlphaSense’s “Sentiment Trend” charts allow you to tap into previous data points and track how management language develops over time.
Each document has an overall “Score,” indicating how “positive” or “negative” a transcript is, and how it compares to the Sentiment score in the previous quarter.
One of the biggest advantages of AlphaSense Sentiment is the ability to rank transcripts by the biggest shift in Sentiment from the previous quarter. This allows you to hone in on the transcripts that you should pay the most attention to. Clicking the delta (Δ) in a list of transcripts sorts the transcripts by the biggest change since the previous quarter.
Comparing scores can be used to gather quick insights about an industry or overall macroeconomic picture.
Sentiment analysis for sell-side research
Sentiment is just one piece of the advanced, AI-powered investment research toolset that the AlphaSense platform provides. When coupled with advanced Synonym recognition, real-time email and push alerts, table exports, Blackline, and a host of collaboration tools, you are again closer to reducing your reliance on previously time-consuming tasks and focusing on the work that really matters.
Jan Svenda is an independent equity analyst focused on the U.S. Small / Micro-cap space where he searches for long ideas trading around Net Current Assets Value (NCAV) and for short ideas which showcase a significant potential for aggressive or manipulative accounting.
More like this
9 min read
10 min read