Most data companies lead with what their product can do. SESAMm's public methodology does that, and then goes a step further. It defines, in plain terms, exactly what the Controversy Exposure Score measures and where its boundaries lie. That precision, now public and free to access under the EU ESG Rating Regulation that entered into force on 2 July 2026, is what makes the score dependable.

The logic is straightforward. A number is only as useful as the user's understanding of it. SESAMm would rather its clients understand the score completely than take it on trust, because a well-understood score is a score that can be used with confidence.

Built From the Public Record

The CES is built entirely from public and licensed media and web content. That foundation gives it a clear and well-defined scope, and SESAMm is precise about what that scope includes.

Coverage is richer for some entities than others. Large and high-profile companies generate far more reporting than small or private ones. SESAMm addresses this directly by rebasing each event against an entity's own media history rather than absolute volume, so a company is measured against its own baseline rather than penalised for simply attracting more press. For entities with a persistently low profile, the methodology is explicit that the underlying signal is thinner, which tells a user precisely where to bring additional sources to bear.

Language and source access define the rest of the scope. The pipeline reads a broad and growing set of languages and ingests an extensive range of public and licensed sources. Where a controversy is reported mainly in a language or a publication outside that set, the methodology says so plainly. Defining these edges is what allows a user to place the score accurately within a wider process.

The Discipline of Not Guessing

One principle deserves particular attention, because it sets SESAMm apart from a common industry habit. Where direct coverage of an entity is thin, SESAMm does not fill the gap with proxy data, sector benchmarks or estimated values.

This is a deliberate quality choice. Substituting averages would produce a tidier-looking dataset, but it would manufacture information that does not exist. SESAMm reports only what the evidence supports. For a low-visibility entity, that means a low score reflects the controversies actually detected, and the methodology is clear that this is a measure of detected exposure rather than a clean bill of health. The result is a number a client can stand behind, because nothing in it is invented.

This alo clarifies how the score is best read. The CES measures exposure to negative controversies, which makes it a sharp, single-purpose instrument. It is designed to surface risk, not to certify virtue, and pairing it with positive-performance data is exactly how SESAMm intends it to be used.

An Early Signal, Drawn From Public Reporting

The CES reflects controversies as reported in public sources, which can include allegations that are still moving through the courts. The methodology is precise about what this means: the score records the existence and salience of reporting, and it is built to give risk teams an early signal rather than a legal conclusion.

This is one of the score's most valuable properties. Reputational and ESG risk very often crystallises long before any legal process concludes. A measure that captures reported exposure as it emerges, while being clear that reporting is not a verdict, gives a risk team time to act early and to weigh the signal appropriately. That combination of timeliness and precision is precisely what makes it useful in practice.

Rigorously Engineered, Openly Documented

Because the score is produced by an AI pipeline, SESAMm documents both how that pipeline works and the controls that keep it accurate. The engineering is the headline here, and it is substantial.

A language-model filter screens for false positives before any event is surfaced. A dual-layer human quality-assurance process, run daily by SESAMm's Research and Analytics team and escalated where needed to the Methodology Lead, reviews accuracy, corrects confirmed issues at the source, and feeds recurring patterns back into the training corpus so the system improves over time. Before any material change to the methodology is deployed, it is backtested against a historical event database, reviewed on the entities it most affects, and signed off by the Methodology Lead. The methodology is formally reviewed at least once a year.

SESAMm also names the structural properties of statistical models openly, because describing a system you understand and control is what gives these safeguards their meaning. The point is not that any model is flawless. It is that the controls are designed for exactly the points where models need them, and that the whole arrangement is documented for anyone to inspect.

Why Defining the Boundaries Builds Trust

There is a quiet truth in ESG data. The providers willing to define the edges of their product are often the ones most worth trusting, because they are describing a system they genuinely understand and operate. A score presented as all-seeing invites misuse. A score presented with its scope clearly marked can be integrated thoughtfully, weighted sensibly, and combined with other inputs exactly as the methodology intends.

SESAMm's view is that this clarity is part of doing AI well, not a step back from it. The same analysts who design the methodology are the ones who test and refine it every day, and the new regulation now gives the whole market a reason to hold itself to the same standard. Precision about what a score means is not a limitation on its value. It is the foundation of it.

‍

To read the full methodology, including how the Controversy Exposure Score is built and the safeguards that keep it accurate, visit sesamm.com/methodology.

Financial and ESG insights begin with big data coupled with data science.

At SESAMm, our artificial intelligence (AI) and natural language processing (NLP) platform analyzes text in billions of web-based articles and messages. It generates investment insights and ESG analysis used in systematic trading, fundamental research, risk management, and sustainability analysis.

This technology enables a more quantitative approach to leveraging the value of web data that is less prone to human bias. It addresses a growing need in public and private investment sectors for robust, timely, and granular sentiment and environment, social, and governance (ESG) data.
This article will outline how the data is derived and illustrate its effectiveness and predictive value.

Content coverage and ESG data collection

The genesis of SESAMm’s process is the high-quality content that comprises its data lake, the source from which it draws its insights. SESAMm scans over four million data sources rigorously selected and curated to maximize coverage of both public and private companies. Three guiding criteria—quality, quantity, and frequency—ensure a consistently high input value.

Every day the system adds millions of articles to the 16 billion already in the data lake, going back to 2008. The coverage is global, with 40% of the sources in English (the U.S. and international) and 60% in multiple languages. The data lake, expanding every month, comprises over 4 million sources, including professional news sites, blogs, social media, and discussion forums.

The following tables illustrate SESAMm’s data lake distribution (Q1 2022):

Respect for personal privacy figures highly in the data gathering process. We don’t capture personal data, like personally identifiable information (PII), and respect all website terms of service and global data handling and privacy laws. SESAMm’s data also doesn’t contain any material non-public information (MNPI).

Deriving financial signals and ESG performance indicators

SESAMm’s new TextReveal® Streams platform applies NLP and AI expertise to process the premium quality content gathered in its data lake. This complex process involves named entity recognition (NER) and disambiguation (NED)—the process of identifying entities and distinguishing like-named entities using contextual analysis—and mapping the complex interrelationships between tens of thousands of public and private entities, connecting companies, products, and brands by supply chain, location, or competitive relationship.

Using SESAMm’s TextReveal Streams, this wealth of information is filtered to focus on four crucial contexts for systematic data processing, risk management, and alpha discovery:

• Sentiment covering major global indices: world equities (and Small Caps, Emerging), U.S. 3000, Europe 600, KOSPI 50, Japan 500, Japan 225
• Sentiment covering all assets and derivatives traded on the Euronext exchange
• Private company sentiment on more than 25,000 private companies
• ESG risks covering 90 major environmental, social, and governance risk categories for the entire company universe, which includes more than 10,000 public and more than 25,000 private companies with worldwide coverage

TextReveal Streams data sets and assessments are used by financial institutions, rating agencies, and the financial services sector, such as hedge funds (quantitative and fundamental) and asset managers, to optimize trade timing and identify new sustainable investment opportunities. Private equity deal and credit teams also use the data for deal sourcing and due diligence. Private equity ESG teams use it to manage initiatives like portfolio company environmental, social, and governance risk and reporting.

Methodology and technology for processing unstructured data

NLP workflow, from data extraction to granular insight aggregation

Data is continually extracted from an expanding universe of over four million sources daily. As it enters the system, it is time-stamped, tagged, indexed, and stored in our data lake to update a point-in-time history extending from 2008 to the present.
The source material is then transformed from raw, unstructured text data into conformed, interconnected, machine-readable data with a precise topic.

Mapping relationships between entities with the Knowledge Graph

At the heart of the text analytics process is SESAMm’s proprietary Knowledge Graph, a vast map connecting and integrating over 70 million related entities and their keywords. It’s essentially a cross-referenced dictionary of keywords, relating each organization to its brands, products, associated executives, names, nicknames, and their exchange identifiers in the case of public companies.

Entities within the Knowledge Graph are updated weekly and tagged to ensure changes are correctly tracked. The CEO of a company today, for example, may not be the CEO tomorrow, and brands may be bought and sold, changing the parent company with each sale. Weekly updates within the Knowledge Graph ensure the system is aware of these changes.

Named entity disambiguation (named entity recognition plus entity linking) is one of the NLP techniques used to identify named entities in text sources using the entities mapped within the Knowledge Graph universe.

At SESAMm, NED identifies named entities based on their context and usage. Text referencing “Elon,” for example, could refer indirectly to Tesla through its CEO or to a university in North Carolina. Only the context allows us to differentiate, and NED considers that context when classifying entities. This method is superior to simple pattern matching, limiting the number of possible matches, requiring frequent manual adjustments, and cannot distinguish homophones.

SESAMm uses three other NLP tools to identify entities and create actionable insights. These are lemmatization, embeddings, and similarity. Each is explained in more detail below.

Analyzing the morphology of words with lemmatization

News articles, blog posts, and social media discussions reference organizations and associated entities in various forms and functions. Lemmatization seeks to standardize these references so the system knows they mean the same thing.

For example, “Tesla,” “his firm,” “the company,” and “it” are all noun phrases that can appear in a single article and refer to a single entity. Even where the reference is apparent, it can take different forms. For example, “Tesla” and “Teslas” both refer to the same entity but have slightly different meanings (semantics) and shapes (morphology).

The lemmatization process standardizes reference shape (morphology) to facilitate identification and aggregation. Lemmatization is a more sophisticated process than stemming, which truncates words to their stem and sometimes deletes information.

Encoding context and meaning with word embedding

In NLP, embedding is a numerical representation of a word that enables its manifold contextual meanings to be calculated relationally. Embeddings are typically real-valued vectors with hundreds of dimensions that encode the contexts in which words appear and, thus, also encode their meanings.
Because they are vectors in a predefined vector space, they can be compared, scaled, added, and subtracted. An example of how this works is that the vector representations of king and queen bear the same relation to each other as the representations of man and woman once you subtract the vector that represents royal.

Using embedding is key to analyzing how words change meaning depending on context and understanding the subtle differences between words that refer to the same concept: synonyms. For example, the words business, company, enterprise, and firm can all refer to the same thing if the context is “organizations.” But they represent different things and even different parts of speech if the context changes.

In the phrase, “[Tesla] will be by far the largest firm by market value ever to join the S&P,” for example, one could replace the word firm with company or enterprise without affecting the meaning significantly. Contrast that with “a firm handshake,” where a similar substitution would render the phrase meaningless.

Also, words referring to the same concept can emphasize slightly different aspects of the concept or imply specific qualities. For example, an enterprise might be assumed to be larger or to have more components than a firm. Embeddings enable machines to make these subtle distinctions.

One advantage of using embedding is that it’s practical because it’s empirically testable. In other words, we can look at actual usage to determine what a word means.

Another advantage is that embeddings are computationally tractable. This understanding of a word’s definition allows us to transform words into computation objects to programmatically examine the contexts in which they appear and, thus, derive their meaning.

As lemmatization is an improvement on stemming, embeddings improve techniques such as one-hot encoding, which is close to the common conception of a definition as a single entry in a dictionary.

SESAMm uses the global vectors for word representation (GloVe) algorithm to generate embeddings. It’s an unsupervised learning algorithm that begins by examining how frequently each word in a text corpus co-occurs with other words in the same corpus. The result is an embedding that encapsulates the word and its context together, allowing SESAMm to identify specific words in a list and different forms of the listed words and unlisted synonyms.

GloVe is an extension of recent approaches to vector representation, combining the global statistics of matrix factorization techniques like latent semantic analysis (LSA) with the local context-based learning of word2vec. The result is an unsupervised algorithm that performs well at capturing meaning and demonstrating it on tasks like calculating analogies and identifying synonyms.

BERT is another algorithm used by SESAMm to generate embeddings. BERT produces word representations that are dynamically informed by the words around them. Google developed the technique, and it’s what’s known as a transformer-based machine learning technique, which means it doesn’t process an input sequence token by token but instead takes the entire sequence as input in one go. This technique is a significant improvement over sequential recurrent neural network (RNN) based models because it can be accelerated by graphics processing units (GPUs).

SESAMm uses BERT for multilingual NLP of its extensive foreign language text because it has been retained using an extensive library of unlabeled data extracted from Wikipedia in over 102 languages. BERT model was trained to predict words from context and next sentence prediction where it was trained to predict if a chosen following sentence was probable or not given the first sentence. As a result of this training process, BERT learned contextual embeddings for words. Due to this comprehensive pre-training, BERT can be finetuned with fewer resources on smaller datasets to optimize its performance on specific tasks.

Linking words, sentences, and topics with cosine similarity

Cosine similarity with centered means it’s identical to the correlation coefficient, which highlights another element of the computational tractability of the embeddings approach. It makes it easy to compare words and contexts for similarity.

Converting words to vector representations means we can quickly and easily compare word similarity by comparing the angle between two vectors. This angle is a function of the projection of one vector onto another. It can identify similar, opposite, or wholly unrelated vectors, which allows us to compute the similarity of the underlying word that the vector represents.

Two vectors aligned in the same orientation will have a similarity measurement of 1, while two orthogonal vectors have a similarity of 0. If two vectors are diametrically opposed, the similarity measurement is -1. In practice, negative similarities are rare, so we clip negative values to 0.

Cosine similarity measures whether two words, sentences, or corpora are close to one another in vector space or “about” the same thing in semantic space. To answer the question, “Is this sentence referencing company X?” we embed the sentence using the process described above and compute the cosine similarity between the sentence and the embedded company profile. Analogously, we compute similarities between sentences and the ESG topics SESAMm monitors by taking the maximum similarity between a sentence and each embedded keyword associated with an ESG topic.

These similarities allow us to identify whether a sentence references fraud, tax avoidance, pollution, or any other ESG risk topic among the more than 90 that SESAMm tracks across the web.

Similarities within ESG topics combine with word counts to resolve the recall and precision problem. Word counts are precise because if a word is identified within a context, then that context, by construction, references the topic.

The virtue of using these NLP techniques is that even if a given keyword list does not include every possible combination of words that a person might use to discuss a topic, relevant entities missed by the word-count process will be identified through vector similarity.

This is the power of SESAMm’s NLP expertise. We can scan many lifetimes’ worth of data in seconds to find the concepts you explicitly ask for and the concepts relevant to your search but that you did not think of yourself.

Sentiment analysis with deep learning and neural networks

Once we’ve identified the concepts and contexts of interest in all the forms they appear, we analyze the context to determine the speakers’ attitudes.

We use sentiment classification models to score a sentence with three possible outcomes: negative, neutral, or positive. The current classification models are based on deep learning AI technologies. Specifically, we stack convolutional neural networks with word embeddings and bayesian optimized hyperparameters—parameters not learned during training. This architecture improves the accuracy and enables fast shipping of production-ready models for a given language. We also produce state-of-the-art frameworks with architecture variations enabling multilingual capabilities, such as transformers and universal sentence encoders.

Condensing information and extracting insights with daily aggregation

Similarities, embedded word counts, and sentiment are state-of-the-art tools for processing unstructured text data. The same tools are effective cross-linguistically.

Once the information has been extracted from millions of data points, it’s aggregated and condensed into actionable insights.

All entities are referenced directly or indirectly within an article. Then, sentence-level references are aggregated to obtain an article-level perspective, and finally, all relevant articles are aggregated to gain an entity-level view of that day.

In this way, reams of data are compressed into several metrics to provide a daily aggregate view for each entity, highlighting trends at a sentence, article, and entity-level comparable over a multi-year history.

ESG analysis use cases

SESAMm’s TextReveal Streams is used in various investment domains, from asset selection to alpha generation and risk management. Systematic hedge funds track retail interest in real time to identify investment opportunities and protect their existing positions. In the Private Equity industry, equity and credit-deal teams use the data in various ways, from monitoring consumer perspectives via forums and customer reviews for evaluating deal prospects to estimating due diligence risks, all to help make investment decisions. Dedicated teams use our data for monitoring portfolio companies for ESG red flags that conventional ESG reporting might miss.

Below are two examples of how aggregated TextReveal Streams data can be used to help identify investment risk and opportunity.

LFIS CapitalL: ESG signals for equity trading

ESG controversies can significantly impact asset prices in the short term, and it’s now estimated that intangible assets, including a company’s ESG rating, account for 90% of its market value.

Working in partnership with LFIS Capital (LFIS), a quantitative asset manager and structured investment solutions provider, SESAMm developed machine learning and NLP algorithms that could analyze ESG keywords in articles, blogs, and social media, to generate a daily ESG score specific to each stock, which is part of the TextReveal Streams’ platform’s core functionality.

The results were promising when these scores were incorporated into a simulated strategy for trading stocks in the Stoxx600 ESG-X index.

A simulated long-only strategy running between 2015 and 2020, using the signals, delivered a 7.9% annualized return, 2.9% higher than the benchmark for similar annualized volatility (17.3% vs. 17.1%). The information ratio of the strategy was greater than 1, with a tracking error of 2.8%. Results for the previous three years were compelling, reflecting the growing interest and news flow around ESG themes.

Researchers also backtested a hypothetical long-short strategy for all stocks in the Stoxx600 ESG-X index with a market cap of over $7.5bn. This investment strategy delivered a Sharpe ratio of approximately 1 with annualized returns and volatility of 6.1% and 5.9%, respectively, between 2015 and 2020. Like the long-only strategy, returns were particularly robust over the three years up to 2020: +6.0% in 2018, +7.3% in 2019, and +11.3% in 2020.

Finally, a simulated “130/30” ESG strategy that combined 100% of the long-only ESG strategy and 30% of the long-short ESG strategy delivered a 10.8% annualized return, 5.8% higher than that of the Stoxx600 ESG-X index. Annualized volatility was similar at 16.9% vs. 17.1%. The strategy experienced a tracking error of 3.8% and an information ratio of over 1.5, with a consistent outperformance each year.

Disclaimer: Past performance is not an indicator of future results. Theoretical calculations are provided for illustrative purposes only. The investment theme illustrations presented herein do not represent transactions currently implemented in any fund or product managed by LFIS.

Read more: “How Alterntive Data Identifies Controversies Before Mainstream Sources With Examples”

Wirecard: ESG sentiment and volume as predictive indicators

The Wirecard scandal broke on June 21, 2020, when newswires carried the story that the major German payment processor had filed for bankruptcy after admitting that €1.9 billion ($2.3 billion) of purported escrow deposits did not exist.

Could SESAMm’s TextReveal Streams platform have provided investors with an early warning that the scandal was about to break?

The following chart derived from the platform shows how key ESG metrics, including ESG scores (volumes) and ESG scores (sentiment), reacted to the news.

An analysis of the charts pinpoints a shallow rise in the ESG scores (volumes) time series in the early part of June before the eruption on June 21.

The ESG scores (sentiment) metric also shows a steady increase in negative sentiment for governance, the most relevant of the three ESG factors regarding the scandal.

Additionally, before the crash, governance was the most negative of the three ESG factors most of the time. This was especially the case from late March to early April, and then before the scandal in early June, negative governance sentiment diverged higher from the other two.

The rate-of-change of negative governance sentiment as it rose and peaked in early June before the scandal broke was also extremely high, perhaps providing the basis for an early warning signal.

Portfolio managers who had been keeping an eye on the reputational slide in Governance for Wirecard may have decided the company was at high risk of a negative controversy emerging, giving them cause to drop the stock before the event.

In this way, it can be seen how while not providing a hard and fast early warning signal, SESAMm’s ESG scores can, nevertheless, be used as the basis for developing a data-driven, rules-based portfolio management approach that can help investors avoid high-risk candidates like Wirecard.

SESAMm takes on ESG data challenges

SESAMm’s NLP and AI tools analyze over four million data sources daily to identify thousands of public and private companies and their related products, brands, identifiers, and nicknames, turning reams of unstructured text into structured and actionable data.

SESAMm’s TextReveal Streams platform can be used in many quantitative, quantamental, and ESG investment use cases. TextReveal is a solution that allows you to fully leverage NLP-driven insights and receive high-quality results through data streams, modular API and dashboard visualization, and signals and alerts.

Learn how SESAMm can support you in your investment decision-making and request a demo today.

To request a demo or for access to the full SESAMm Wirecard or LFIS reports, contact us here:

Controversial business involvement is no longer a niche ESG issue. From fossil fuels and weapons to gambling, sanctions, and human rights abuses, exclusion policies are expanding and scrutiny is intensifying, especially across private markets.

As SFDR and the EU Taxonomy raise the bar, investors must prove that controversial exposures are identified, documented, and consistently screened, even when disclosures are limited.

In this ebook, SESAMm explores:

• How exclusion rules are evolving under SFDR, the EU Taxonomy, and investor mandates
• Why controversial involvement is harder to detect in private markets and secondaries
• Real-world case studies revealing hidden exposure and compliance risk
• How AI enables faster, auditable screening across public and private assets

Download the ebook to learn how investors can apply consistent, defensible exclusion screening at scale.