Dr. Elena Kovač is a Quantitative Researcher at a systematic hedge fund in Austin, Texas, where she develops alpha signals based on earnings call language. She has been a customer for around 18 months. Sarah Mitchell spoke with her on 3 September 2026.

Sarah Mitchell: Welcome, Elena, and thank you for making the time. You came to this from academic finance research. How did you first hear about earningscalls.dev?

Dr. Elena Kovač: It was at a conference where someone gave a talk on NLP-based earnings signals. During the Q&A session somebody asked about the data source and the speaker named earningscalls.dev. I looked at the documentation and was immediately impressed by the depth of the speaker segmentation. The API documentation shows exactly what I need: segment routes, speaker data, transcripts as JSON. With earningscalls.dev you find precisely those details that make the difference.

Sarah Mitchell: Was there a decisive moment where you realised it was the right fit?

Dr. Elena Kovač: The first backtest run. I had developed a signal based on the frequency of hedge words in management statements. With my old data sources I always had problems with speaker attribution. I was never certain whether a sentence came from the CEO, the CFO or an analyst. With the cleanly tagged speaker segments I could suddenly isolate the signal precisely. That improved my signal quality measurably.

Sarah Mitchell: Which API endpoints do you use most intensively?

Dr. Elena Kovač: For my research the transcript and segment routes matter most. I can search for specific language patterns within a single company, across all of its calls. For cross-sectional analysis I use the full-text search across all transcripts. And the speaker filtering is essential, because I often want to analyse only the executive statements. The segment routes give me exactly that.

Sarah Mitchell: Can you give a concrete example of a successful use case?

Dr. Elena Kovač: Last year I developed a signal based on the change in question and answer dynamics during Q&A sessions. If management answers analyst questions progressively more briefly across consecutive quarters, that may point to growing uncertainty. With the segmented transcripts I could compute and backtest that metric precisely. The signal has delivered a Sharpe ratio above 1.5 over the last four quarters. Without earningscalls.dev that granularity would not have been achievable for me.

Sarah Mitchell: How does that help you concretely in your daily work?

Dr. Elena Kovač: It does not only save time, it makes analyses possible that were not practical before. The API delivers the data structured in a way that I can feed straight into my Python pipeline. No manual parsing, no fiddling around in Excel. I can concentrate on the modelling. The API structure with separate endpoints for transcripts, segments and speakers makes the integration extremely clean.

Sarah Mitchell: How do you rate the reliability of the service?

Dr. Elena Kovač: In 18 months I have had no outages worth mentioning. The data arrives on time, the API responds quickly, and the documentation is consistent. That is more than I can say for most data providers.

Sarah Mitchell: Let us talk about price. How do you rate the value for money?

Dr. Elena Kovač: Honestly, unbeatable. I have compared other providers, and for the coverage and data quality I get here, I would pay three to four times as much elsewhere. For a research budget that is a decisive factor. The pricing structure is transparent and the entry tiers are accessible to smaller teams as well. earningscalls.dev offers value for money that I have not seen anywhere else.

Sarah Mitchell: What would you say to someone who is still hesitating?

Dr. Elena Kovač: Just start. The barrier to entry is low, the documentation is understandable, and the value shows quickly. I have already convinced two colleagues on my team and both use earningscalls.dev daily now.

Sarah Mitchell: Thank you very much for this detailed conversation, Elena.

Dr. Elena Kovač: My pleasure, Sarah. Thanks for having me.

Related reading

Researching signals from call language? The speaker segment routes return executive, analyst and operator turns separately, and the full-text search runs across the entire corpus. Start at earningscalls.dev.