Automatically Analyze Public Company Earnings Calls With Transcription & Natural Language Processing

This is is part of my live-learning series! I will be updating this post as I continue through my journey. I apologize for any grammatical errors or incoherent thoughts. This is a practice to help me share things that are valuable without falling apart from the pressure of perfection. 

Speak With Tyler Bryden
Automatically Analyze Public Company Earnings Calls With Transcription & Natural Language Processing
  • Rapid prototype for automatic analysis of earnings calls
  • Extracting valuable keywords and sentiment
  • Scraping earnings calls recordings and transcripts
  • Findings from other NLP research on earnings calls
  • Sample output with Tesla Q1 2022 earnings call
  • Request for feedback from connections who invest

#earningscalls #tesla #NLP #sentimentanalysis #keywordextraction #namedentityrecognition #financialanalysis #transcription #speechtottext #machinelearning #artificialintelligence #buildinpublic

Resources (password is earnings)

YouTube Video


Automated Transcription

Hello hello Tyler Bryden here. Interesting rabbit hole. I ended up down yesterday what I was doing was putting together a page page was for about basically about transcribing an earnings call and one of the reasons why I mean what what we’re doing on speaker. I continue to build out. Basically these templated well SEO optimized optimized pages and we’re taking sort of the the how to the actual method that you’re trying to do. So for example, in this one transcribe and then the actual data source that you would be applying. And technology to when using speak for. So in this case an earnings call.

And so built a little guide, you know. Nice to walks, walks through everything, and one of the triggers to do this was that we’re talking to our transcriber team who fulfills all the professional transcription that we do, and then obviously has a bunch of clients that they’re sourcing. They do an incredible job, and they said they were just overloaded right now because they were doing all the quarter one 2022 earnings calls. They were transcribing them professionally because they are sort of responsible for doing. You don’t. It’s not mandated, but most companies will do an earnings call. CEO or executives will then share insights from the actual from the actual quarter. Go over, go over the the performance, take questions. Sometimes there’s things that come through those questions from analysts who are in the room asking them those questions, and then sometimes there’s some off the cuff moments that can reveal a lot of information, and so they are after they have that conversation, they are responsible for them doing a professional.

Fully accurate transcript that then is published on the site and so find lots of those available there was that for just as a quick example to make this more clear was Applied Materials. They did their earnings calls and then here is the actual PDF that then was published so they have it a little bit more formatted. I’ve seen Larry the ones that are just raw transcripts here, but it’s a really sort of interesting piece that sort of emerged. And so I said, well, this is worth it. Investigating a little bit more ended up creating this page and this is a template page that then we do for different outputs and then I was like you know what? What does this look like to? If I apply this transcription, but then the automatic analysis of speak that that we can do and I then start like? Is this even worth doing? Is there something you know worthwhile here and so I did a little bit. It looks like do people actually watch earnings college? They listen to them? Is there any insights that that come from this and it was some really interesting pieces that came out which was like?

The importance of earnings cause it’s acknowledged and that investors really they do. They plan to close their trades. They plan to trades, do their trades closer to the date of an upcoming conference, and then they often use that to like update their earnings estimates and then make make better decisions. And then there’s also a lot of people who are listening to these calls who are sitting on the sidelines. As in they’re not investors yet, but they’re deciding if they want to invest in this company and so they listen to these calls. Try to figure out if it’s they sound confident, if they sound optimistic, if there’s any insights into what happened in that previous quarter, and there’s a lot of signals of moving forward, and even this idea of a guidance. And there can be good guidance. Or there can be, you know, not so good guidance that comes out and then through some of these questions. Maybe there’s information that is revealed.

That’s not quite as there wasn’t necessarily supposed to be revealed, so some really interesting moments that can happen there, and one of them. One of the just an example of that, was a moment in one of our Teslas. Things where Elon was obviously frustrated, annoyed by questions and did not, you know, sounded a little bit pessimistic and and that had an impact on on then the trade volume and and price, and so the actual earnings calls are these micro moments that have an impact on the larger market and even you know a couple percent or even less than a percent. That’s millions if not billions of dollars. But then it’s moving around for the economy, so there’s a pretty big impact there. And you know, talk about headwinds, tailwinds, all this, all this stuff so.

There have been some really fascinating studies that have been done where they’ve scraped, say, historical data sets of transcripts and then looked at the performance. See if they could find any correlation, or at least connection between those data points and there has been some really interesting ones, one including the difference between a morning call and an afternoon call, and if something gets pushed back to the afternoon, there seems to be a little bit more struggle during that call, and there’s actually been sort of this correlation. Our connection, at least with a dip in about 1.5% stock price dip on the year, so some really fascinating things that have come out of this just from like some some surface level research some you know, some GitHub library looking and you know. Again, ended up down the rabbit hole and having a lot of having a lot of fun. And sometimes you don’t have fun fun. Fun is important when you’re when you’re doing all that stuff and then just a couple last ones. That sort of stuck out so it was a city.

Firms that reference more positive descriptors around their financials, they outperformed historically and then firms that were more transparent in these calls, they also outperformed historically. So a couple really interesting things, at least worthwhile for me to figure out if I could investigate a bit more. And so I did. And So what I did was took a Tesla call and I’ll share. I’ll share this and you can see on the site here. Tesla earnings call and then I’ll talk of this a little bit more. But really there’s an iframe player player that’s there. A couple highlights from it. And then the entire raw transcript from us for this perspective. When people are searching for this, there’s an SEO benefit to it, and transcripts are always good for SEO, collapsible accordion, and then a couple sort of high level.

Insights and then in this case these these pieces are clickable, so you can actually click through these specific moments and then jump in here about growth and this earnings call. So from a navigation or a streamlined perspective to get to the moments that matter, there’s something really powerful there and this is where I think speak starts to shine a little bit, especially if you are a non technical person, which is I looked into. OK, there’s a couple of research papers around the keywords and phrases or then just by scraping a couple seeing that there are language that are used. Commonly in these calls that then are important moments for people who are investors or prospective investors to understand and then take note of and. So this is the higher level group. Things that come from earnings calls and you can see a little bit of, you know some some interesting terms here are around, you know free cash gift talking about COVID, inflation etcetera etcetera. And then you can use the custom categories to then break this down even further. So instead of just having it all grouped as earnings call you start to build subcategories.

One of those big ones, forward looking statements and then as we analyze these these calls realize there’s a lot of forward looking statements in these, so you can’t take these without utmost. Belief, and this is where I find there’s the separation between earnings calls, sort of financial or sort of financial earnings, which are historically looking at the the previous quarter, whereas in these earnings call case they’re actually projecting and looking forward into the future. So there’s some very interesting insights, and you can see when these forward looking statements emerge and then things around revenue, etcetera etcetera. I just want to see here like what I can do then. Like very simply, I can say, you know.

You know, just as an example, just type the word fast automatically. That is now added to my earnings call piece. Here I can jump into one of the earnings calls that I did, so this was that Tesla one that I was saying and you can see that everything’s there and I can then hit reanalyze and if the word fast is mentioned throughout then it will automatically be picked up from them. So this was really a fascinating to me. There is the idea of like identifying speakers, but then you can see all the mentions of forward looking statements. I think if I.

You know, jump here and I say, expect.

But I’m I. I feel very high confidence in the teams of both factories and.

But there you go, you’ve got a mention of a forward looking statement and what we should be doing a little bit better here is just displaying this in sentence format. So very quickly you could get almost a summary or a report of this earnings call without necessarily having to listen to the entire thing, and so this is a really fascinating idea to me. First of all, I’m interested in this. We’re in a volatile market right now. There’s a lot going on and I believe that there are other people who are interested in, especially ones who have lost a lot of money or actually have money. Right now and are looking to invest it as this quarter. I mean, a lot of these earnings have been done here. Now for Q one 2022, but the next quarter as we look at where are we in this economy? What is this recession look like? Yes, what does this recession look like? Because we’re in a recession. These earnings calls might be very important and impactful for how we then project and move forward. So definitely some interesting pieces around here. What I you know, what was you know what then happens here? As you can build these little databases of OK? We’ve got Tesla earnings calls and then you can use speak. You can see them nicely organized here and then. There’s obviously some nice call to actions and stuff, so there’s something really sort of interesting about this and.

Why this was fascinating to me. Just to show you a little bit of my thought process. Or you know why. I sort of spent a little bit of time on it was I took typed up earnings calls about total volume of so this is SEM rush, SEM rush. I don’t know how you pronounce it, which really sort of finds the search volume. So volume the trend if it’s trending up or down. How difficult is it to rank for these terms? How much would it be for a cost per click? And then you can even see the results page of this and so you can see. Here’s all the people who are looking for earnings.

Balls and then the volume that then is associated with each one of them and and then using this. I could now say, OK, you know we’re going to start to dump a couple earnings calls, transcripts in, and then we’re also going to produce this automatic analysis now that this custom categories is ready and I can then start to get the search engine rankings, I can start to drive people to the site. They can get the insights directly from speak. They can be familiar with speak, and might even have their own calls that are doing their own webinars, their own financial presentations. So a lot of interesting things that can just come out of a small piece because of how easy it is to customize the insights within the speak system.

And you know more than just the ones that I automatically programmed in. Speak is obviously doing some extraction of people of brands, and then even things like dates, Cardinals, geopolitical regions, money so 5.

100 that might be like, you know.

Etcetera etcetera and then product so we know that he was talking about the Model 3 jump to that moment. So a lot of power with the combination of the named entity recognition, the custom categories, and then the navigation element that speak has. There are some challenges with accuracy of transcripts, and I think it’s really important to do it. Right here, but generally this is good quality audio and the transcription is is strong. The other option here, which is pretty easy, is I could actually just go. You know, like this one. I’ve got great. I’ve got an actual PDF here. I jump in, I hit new upload and then I grab let’s see here.

PDF push it in. Upload and it’s going to analyze this PDF right away so it’ll take a second here just to analyze everything. I’ll refresh in a moment and then and then we’ll see. Here’s how that we here’s how the this transcript that was more professionally published. And then what that accurate is there. So again, you can see forward looking statements the all the expect, and then believe. And then even if we want to look at OK, talk about free cash, incremental, free cash flows, so some really fascinating.

Things that I gained from this perspective obviously good to know when you are a business you know owner or you’re an investor. You’re paying attention to the markets and it’s having an impact on your life. So definitely doing some stuff there and then. The other thing that I just wanted to show quickly, which is what will become, I think really fascinating is. I can then share this into this library. So I’ve got if I remember my own password. There we go earnings now I have all time I can look at.

The different sort of automatically visualized. So this is now the aggregated analysis. If I then hop into this and I move it to folder, earnings calls update all. So now this would be in there too if I go hop I refresh this we can see it included. Everything can be tagged by the quarter, so I know that I’m only interested in Tesla for example. I’m interested in audio video. I can apply this filter. We can specifically see that and then they can even jump into the actual categories itself so.

A lot of stuff sort of came out this very quickly. If I want I can say OK, I’m interested in cost. Here is the exact mention of cost and then if I I can want this is that I think I think specific moment. So again a lot of power from this navigation element and just a little interesting deep dive that I had. So I’d be interested to know if you’re investing you’re listening to earnings call, you’re just an investor in general. Is this interesting to you? Is there any feedback that have? Is there anything that you’re looking for from the earnings calls that?

That are not included or that you would like to see included. How do you monitor these? How do you get your insights from this? How do you derive value from these earnings calls? What do you learn? How does that connect with the actual financial performance that you’re looking for? And then how does that all culminate together to make to help you make hopefully better decisions when you’re investing in the part that I’m fascinated with, this is, you know, these impact. There’s impact from the public markets that then rained down on the private market. So as a company that is private market company that has been impacted by.

What’s happening in the world. The volatility of the market. The chance for recession slash that we’re in a recession. All of this is connected and more connected than I had previously thought. So the idea of both the SEO opportunity, the automatic analysis opportunity, and how easy it is to program it to speak, and the ways that we could create value for people. Some of these people, being investors, whether large investors or even small investors, paying attention to the markets and helping them summarize earnings calls and giving them insights as quickly as close to real time as possible so they can make better decisions.

Seems like there’s some value there. Would love to explore it and would love to hear from you. If you have any thoughts, so I hope this was interesting. Thank you so much for checking this out. I know this was a bit of a deep dive and and I was walking through speak which I don’t always try to do, but so it’s fun. I hope you had a hope you enjoy it and I hope you have a great rest of the day. Thank you so much. Bye bye.


More To Explore

Share This Post

Join My Personal Newsletter ❤

Get insights and resources into awareness, well-being, productivity, technology, psychedelics and more.

Don't want to chat but want to keep updated?

I'd love if you subscribed today. I promise I will only send you great, valuable content that has transformed me and helped others flourish. 

You have Successfully Subscribed!

Pin It on Pinterest