The Qualex Origin Story
About a decade ago, I was watching the presidential primary debates when a graphic appeared on the screen showing the grade level at which each candidate spoke. The political analysts noted how Candidate A spoke at a 10th-grade level, Candidate B at an 8th-grade level, Candidate C at an 11th-grade level, and so on. That moment proved pivotal in shaping my research interests and ultimately my career.
The Epiphany
The idea, subject to further investigation, was the possibility of using a person’s speech to quantify their intelligence, perhaps even assigning an IQ-like score. As a fundamental equity analyst, I immediately considered the applications in the stock market. Would it be possible to take quarterly conference call transcripts, apply natural language processing techniques, and rank-order the smartest CEOs and CFOs in the market? Would those scores correlate with future stock price performance? Do quantifiably smart management teams run better businesses and ultimately drive above-market shareholder returns? What other insights might be derived from the free speech of management teams?
Vocabulary as a Proxy for Intelligence
What followed was an exploration of the academic research on intelligence measurement. As far back as 1916, Stanford professor Lewis Terman, a pioneer of IQ testing, found that vocabulary strength was highly correlated with overall intelligence. Similarly, Robert Sternberg of Yale University wrote in his 1987 book:
“Vocabulary is probably the best single indicator of a person’s overall level of intelligence… if one wants a quick and not-too-dirty measure of a person’s psychometrically measured intelligence… vocabulary is generally the best predictor of overall score on a psychometric IQ test.”
These and other studies confirmed the basic premise that reliable cognitive insights could indeed be extracted from speech.
Proof of Concept
Quantitative backtests were then performed using many of the metrics common to the field of computational linguistics. Proxies for vocabulary strength included well known readability measures; token-level measures such as average word length and syllable counts; and corpus-level metrics such as lexical diversity and various bag-of-words approaches. Several of these basic linguistic metrics performed well in market and industry backtests, providing the proof-of-concept I was seeking at the time.
Times and Seasons
Around this same period, my wife and I decided to move from Boston to Colorado to be closer to family. Our young family was growing, and the lives of our four children were becoming increasingly busy. We had a strong desire to be present and involved during their formative years. The intelligence research that so intrigued me was left behind and nearly unreferenced for the next seven years.
Artificial Intelligence, Optimizations, and Golf
During that time, however, I joined one of the fastest-growing business segment at my company and became involved in scaled data workflows, data automation, and optimization mathematics within a portfolio construction framework. Optimizers, as many know, are central to artificial intelligence algorithms and model training. This experience provided a valuable foundation for understanding and applying the rapid advancements unfolding across the AI landscape.
Along the way, I took on a side project that applied my optimization and coding skills to a passion of mine - golf. I developed the mathematical methodology and data backbone for a golf ranking / stats SaaS start-up. The matrix optimization ranking approach was able to predict future tournament outcomes with greater accuracy than officially used professional golf ranking systems. After several years of refinement, the platform was acquired by a collegiate golf recruiting organization.
Pursuing the Opportunity
By early 2025, the desire to pursue untapped opportunities in intelligence research returned. Advances in compute power, AI tools, and large language models were opening new doors in the NLP space. I made the decision to leave a nearly 19-year career with one of the premier asset managers in the country to pursue the opportunity ahead.
Research Breakthroughs
Over the next ~15 months, I was immersed in both academic linguistics research as well as proprietary market backtesting. On the academic side, the evidence continued to strengthen that speech patterns are strongly linked to intelligence. For example, a study published in Brain Sciences found that researchers could detect the likelihood of someone developing dementia more than a decade before clinical diagnosis simply by analyzing the pronoun-to-noun ratio in a patient's speech. As brain function and memory decline, individuals unconsciously compensate by substituting vague pronouns such as “it,” “he,” “she,” or “thing” in place of precise nouns.
In another study, the grammatical patterns used in college admissions essays were predictive of future academic performance. Among several findings, higher grades were correlated with greater usage of prepositions and articles, words that function as the structural scaffolding for complex thought.
Intelligence in Equity Markets
On the primary research side, I refined how these concepts applied to public markets. It became clear that the goal was not merely to identify CEOs who speak with the greatest complexity or largest vocabularies. Often, the most effective leaders are those who can distill complex ideas into clear, understandable, and actionable messages.
As a fundamental equity analyst and portfolio manager, I observed this repeatedly. The best managers take complicated situations, identify elegant solutions, and mobilize large organizations around a clear vision. An element of practical intelligence is required to run a company well. Often, it is the individual with the right combination of traditional cognitive ability, emotional intelligence, and practical judgment who drives superior shareholder returns. Luckily, a person's personality, management style, and strategy, among other attributes, can also be extracted from their speech patterns.
This reality sharpened the focus toward identifying linguistic metrics that capture efficiency of speech, logical structure, and clarity of communication. Those themes consistently tested well in market backtests. Over time, the process evolved into a systematic framework capable of rank-ordering management teams on a multi-faceted intelligence scale.
Qualex Investment Management
This research ultimately formed the foundation for launching Qualex Investment Management, a firm focused on Quantifying the Alpha in CEO Lexicon. The fund seeks to invest in the most intelligent management teams in the market and short those on the opposite end of the spectrum. A unique, systematic, research-backed approach that targets S&P 500 outperformance over the long term.
References:
-
Terman, L. M. (1916). The measurement of intelligence: An explanation of and a complete guide for the use of the Stanford revision and extension of the Binet-Simon Intelligence Scale. Houghton Mifflin.
-
Sternberg, R. J. (1987). Most vocabulary is learned from context. In M. G. McKeown & M. E. Curtis (Eds.), The nature of vocabulary acquisition (pp. 89–105). Lawrence Erlbaum Associates.
-
Bittner D, Frankenberg C, Schröder J. Changes in Pronoun Use a Decade before Clinical Diagnosis of Alzheimer's Dementia-Linguistic Contexts Suggest Problems in Perspective-Taking. Brain Sci. 2022 Jan 17;12(1):121. doi: 10.3390/brainsci12010121. PMID: 35053864; PMCID: PMC8773561.
-
Pennebaker JW, Chung CK, Frazee J, Lavergne GM, Beaver DI (2014) When Small Words Foretell Academic Success: The Case of College Admissions Essays. PLoS ONE 9(12): e115844. doi:10.1371/journal.pone.0115844