LSEG MarketPsych Transcript Analytics

Corporate calls, especially earnings conference calls, include unique C-level comments and equity analyst reactions that are pivotal in shaping stock valuations. However, traditional methods of dissecting these conversations are often time-consuming and subjectively interpreted.

Our Solution

Access LSEG MarketPsych Transcript Analytics: a refined data feed that leverages cutting-edge Natural Language Processing (NLP) to decipher corporate discourse quickly and systematically.

This solution delves into the nuances of every spoken sentence, offering insights into sentiment, relevant companies and topics being mentioned, and the intricate web of emotions speakers convey. Designed for the discerning eyes of institutional and quantitative firms, LSEG MarketPsych Transcript Analytics provides a detailed, data-driven perspective on the undercurrents of corporate calls.

By the numbers

Coverage: 16,000+ global public companies

Data frequency: Event-driven

Delivery: API (JSON and CSV outputs)

Document types: Earnings conference calls, conference presentations, guidance calls, sales presentations, and more

Granularity: From sentence-by-sentence up to document-level feeds

History: 2001 onwards

Source: LSEG Transcripts

Versions: Standard and Premium

Dataset Packages

Available in Standard and Premium

This analytics product is available in two versions: Standard and Premium. Both versions offer the full range of entity, topic, sentiment and emotional scores.

The two product versions, Standard and Premium, are structured according to the level of aggregation. Users interested in an overview of the company discourse may opt for the lighter Standard version, while users who require more detail should choose Premium.

Both versions are accessible via API, allowing querying for detailed features, including date range, mentions of certain entities or topics, and free text search. Additionally, both versions are supported by a user interface, as depicted in Figure 5.

Example of the user interface accompanying the Premium version. In the example, all sentences in Tesla’s earnings conference calls containing words such as ‘FSD’ and ‘self-driving’ are counted per quarter (blue bars). The sentiment of each sentence is averaged (red line) and plotted, while the text is displayed in the table below.

Client Applications

Improve fundamental analysis

Fast-track the review of transcripts at scale to capture a more complete landscape. Extend your view beyond standard tabular financial data.

Figure: Intel Corp’s earnings conference call sentiment and stock price. Earnings call sentiment provides additional dimension beyond financial metrics.

Client Applications

Maximise risk management

Identify emerging signals in companies’ outlooks by viewing how companies increase or decrease mentions and sentiment around certain topics and entities.

Figure: Elon Musk’s most mentioned topics (size of the circles) during Tesla’s earnings conference calls. The x-axis indicates optimism (future-oriented sentiment), while the y-axis indicates financial sentiment. Musk spoke the most about Tesla’s several products; displayed the highest optimism in sentences mentioning the topic of “demand”; and the highest sentiment when referencing “software”.

Client Applications

Optimise quantitative strategies

Integrate transcript signals into systematic strategies. Combine topic and event filters with numerical scores to generate thousands of features for quantitative models.

Figure: Average next-month return from 2003 to 2024 of portfolios created by the ranking of US companies according to the level of sentiment in their latest earnings call transcript. The stock performance of companies that show up in the top (bottom) 10% of sentiment tend to outperform (underperform) their peers in other quantiles.