Earnings season already started with $AAPL offering a dim light in an otherwise "dark earnings season".

We decided to run a project and develop a Python script able to calculate the tone of the earnings call by searching for both positive and negative words in the call transcript.

In 2011, a white paper by Loughran-McDonald showed that word lists developed for other disciplines misclassify common words in financial text. In a large sample of 10-Ks during 1994 to 2008, almost three-fourths of the words identified as negative by the widely used Harvard Dictionary are words typically not considered negative in financial contexts.

Hence, we used the Loughran-McDonald Master Dictionary to analyze earnings call transcripts. The code we developed first look for positive and negative words (with a negation check), computes the ratio of positive words over negative and finally generates word clouds.

The code is open-sourced and available on our GitHub: https://github.com/SLTResearch/earnings_call_analysis

To test the relevancy of our tool we decided to run a test on $GOOGL and compare the tone of its earnings calls between Q4 2021 and the most recent one, Q3 2022.

The algorithm spotted 160 positive words vs. 34 negative ones in the Q4 2021 earnings call transcript resulting in a ratio of 4.71. The stock surged +7.52% on the news.

Positive:

Negative:

Now focusing on the latest earnings call (Q3 2022), the algorithm spotted 130 positive words c.20% less) vs. 40 negative ones (c.18% more) giving a ratio of 3.25. The stock dropped -9.14% on the news.

Positive:

Negative:

It is interesting to compare word clouds of the two periods were words like great, strong, challenging and loss are significant in both periods. However, in the most recent one we can see that opportunities is more dominant and on the negative side, challenges, decline and slow(down) are taking over words like threats and vulnerabilities as if the latter have been now materialized.