Paper: arXiv 2310.04280

Authors: Maksim Papenkov, Chris Meredith, Claire Noel, Jai Padalkar, Temple Hendrickson, Daniel Nitiutomo, Thomas Farrell

Abstract

Accurate industry classification is a critical tool for many asset management applications. While the current industry gold-standard GICS (Global Industry Classification Standard) has proven to be reliable and robust in many settings, it has limitations that cannot be ignored. Fundamentally, GICS is a single-industry model, in which every firm is assigned to exactly one group - regardless of how diversified that firm may be. This approach breaks down for large conglomerates like Amazon, which have risk exposure spread out across multiple sectors. We attempt to overcome these limitations by developing MIS (Multi-Industry Simplex), a probabilistic model that can flexibly assign a firm to as many industries as can be supported by the data. In particular, we utilize topic modeling, an natural language processing approach that utilizes business descriptions to extract and identify corresponding industries. Each identified industry comes with a relevance probability, allowing for high interpretability and easy auditing, circumventing the black-box nature of alternative machine learning approaches. We describe this model in detail and provide two use-cases that are relevant to asset management - thematic portfolios and nearest neighbor identification. While our approach has limitations of its own, we demonstrate the viability of probabilistic industry classification and hope to inspire future research in this field.

Complexity vs Empirical Score

  • Math Complexity: 6.0/10
  • Empirical Rigor: 4.5/10
  • Quadrant: Lab Rats — high math complexity, limited empirical testing

Why this score: The paper employs probabilistic topic modeling with Latent Dirichlet Allocation, involving Bayesian concepts and probabilistic derivations, but relies heavily on conceptual explanation and qualitative validation, lacking the backtested performance metrics and code often found in empirical work.

Research Flowchart

  flowchart TD
  A["Research Goal: Develop a probabilistic model for multi-industry classification to address GICS limitations"] --> B["Data Input: Company business descriptions"]
  B --> C["Methodology: Apply NLP Topic Modeling"]
  C --> D{"Computational Process: Assign probability weights per industry"}
  D --> E["Outcome 1: Thematic Portfolios construction"]
  D --> F["Outcome 2: Nearest Neighbor identification"]
  E --> G(("Final: Interpretable Multi-Industry Simplex MIS Model"))
  F --> G

Evidence Card

What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card

Proposes

The paper introduces Multi-Industry Simplex (MIS), a probabilistic extension of GICS that assigns firms to multiple industries with relevance probabilities.

Source sentence
We attempt to overcome these limitations by developing MIS (Multi-Industry Simplex), a probabilistic model that can flexibly assign a firm to as many industries as can be supported by the data. 1.3 New Solution (and its Value Proposition) We introduce the Multi-Industry Simplex (MIS) as a probabilistic extension of GICS.

Assumes 2

  • All source documents are comprehensive and proportionately-representative of a firm's business ventures. data
    Source sentence
    Fundamental Data Assumptions : all documents are comprehensive (they cover all non-trivial business ventures of a firm) and are proportionately-representative (business-ventures are mentioned in proportion to their relevance to a firm; if something is discussed in greater detail - it must have higher relevance).
  • Industry identification relies on a bag‑of‑words representation where word frequencies are sufficient and word order is irrelevant. model
    Source sentence
    3.2 Bag-of-Words Representation (BOW) In principal, MIS identifies industries by counting keywords - which thus requires keywords to be easily extractable from our text.

Data

  • 10‑K Business Descriptions text-news unknown
    Source sentence
    Examples of appropriate texts include : • 10-K Business Descriptions : self-disclosed descriptions of high-relevance business ventures. • Analyst-Written Business Descriptions : third-party descriptions of business ventures. • Earnings Call Transcripts : discussions on existing and emerging ventures of a firm.
  • Analyst‑Written Business Descriptions text-news unknown
    Source sentence
    Examples of appropriate texts include : • 10-K Business Descriptions : self-disclosed descriptions of high-relevance business ventures. • Analyst-Written Business Descriptions : third-party descriptions of business ventures. • Earnings Call Transcripts : discussions on existing and emerging ventures of a firm.
  • Earnings Call Transcripts text-news unknown
    Source sentence
    Examples of appropriate texts include : • 10-K Business Descriptions : self-disclosed descriptions of high-relevance business ventures. • Analyst-Written Business Descriptions : third-party descriptions of business ventures. • Earnings Call Transcripts : discussions on existing and emerging ventures of a firm.
  • News Headlines text-news unknown
    Source sentence
    Beyond this, we can also supplement our dataset with news headlines and social media feeds - though such data sources provide less structure and require additional considerations for stable integration.
  • Social Media Feeds alternative unknown
    Source sentence
    Beyond this, we can also supplement our dataset with news headlines and social media feeds - though such data sources provide less structure and require additional considerations for stable integration.

Costs

Not applicable The paper reports no trading or execution result.

Finds 2

  • MIS is presented as a proof‑of‑concept probabilistic industry classification tool, demonstrating viability despite the inability to perform a historical backtest. theoretical
    Source sentence
    While our approach has limitations of its own, we demonstrate the viability of probabilistic industry classification and hope to inspire future research in this field. The particular MIS model instance that we show in this paper uses a large collection of private and public text datasets that are not all available before 2021, and because of this a proper historical backtest is not possible.
  • Real‑estate and REIT firms were removed from the sample due to instability caused by semantic ambiguity and high false‑positive rates. negative
    Source sentence
    2Due to observed instability, we elected to remove all real estate and REIT firms from our sample. This instability is primarily attributed to semantic ambiguity in the business descriptions, and exceedingly high quantities of false-positives in the model.

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 8 related papers). Coverage is partial, so this is neither replication nor refutation.

Cited by 2 papers in the archive (1 by independent authors)

Code

No repository by the authors found in the paper.

Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.