The Truflation Deadlift Index is the kind of thing that sounds like a joke for about a paragraph and then stops sounding like one. It was built on Pieter Levels' observation that CEOs who lift heavy outperform their peers on the S&P 500 — a finding most people receive as charming until they look at the correlation, at which point it becomes interesting. The index does not claim that deadlifting causes good leadership. It claims that physical capability is a proxy for something deeper — discipline, energy management, the willingness to do uncomfortable work daily, the absence of the small surrenders that compound over time. The deadlift is the visible part of an iceberg the spreadsheet does not see.

It also opens a category. Once one quirky CEO-behavioural index is on the same feed as the conventional metrics, the question of what else might belong there becomes serious. What other non-financial proxies, tracked at population scale, capture something the official numbers miss?

The cleanest candidate, and the one that pairs most naturally with the Deadlift Index without crowding it, is psychological rather than physical. It would index leadership character through a single, mechanically trackable signal in the language CEOs use about their own companies. To stop this from being purely theoretical, the piece also sketches the first cohort — twenty CEOs from the same broad pool the Deadlift Index draws on — at the end.

The 'I-We' Index.

The hypothesis is straightforward. Take the ratio of first-person-singular pronouns — "I", "my", "me" — to first-person-plural pronouns — "we", "our", "us" — used by a CEO in earnings calls and shareholder letters. Calculate it quarterly per company. Aggregate across the index. The number that results would track something revenue, margins and ROE do not: how the person at the top is positioning themselves with respect to the institution they are running. A high "I" share is the linguistic shape of ego-centred leadership. A higher "we" share is the linguistic shape of leaders who locate themselves inside the team rather than above it. The ratio is a single number per CEO per quarter — easy to plot, easy to compare, easy to act on.

Why pronouns aren't whimsy

This is where most quirky-index proposals collapse — they are pleasant intuitions without a research bed underneath them. The 'I-We' Index does not have that problem. The underlying science is forty years of work by James Pennebaker at the University of Texas, who has established across a long series of studies that pronoun use is one of the most stable and most predictive verbal signatures a person produces, and that the I/we ratio in particular tracks closely with leadership effectiveness, team health, narcissism markers and decision quality.

The finding that matters most for an index: leaders who use more first-person-plural pronouns are not, as the lay intuition might guess, the modest or humble ones strategically avoiding "I" for performance. They are typically leaders who genuinely think in collective terms, share credit, distribute decision-making, and run organisations that outperform on retention, output and crisis response. Leaders who use more first-person-singular pronouns — particularly under stress — display the pattern Pennebaker has called narcissistic linguistic dominance, which correlates with ego-bottlenecked organisations, higher executive turnover beneath them, and worse long-run performance.

This is not a soft correlation. It is a robust finding that has filtered into clinical psychology, organisational research and the political-speech literature. The reason it has not yet been turned into a tradeable index is not that the signal is too weak. It is that nobody has bothered to build one.

Why it works as a tradeable index

The methodological case for the 'I-We' Index is unusually clean for a quirky index. Three reasons hold it together.

First, the data already exists. Every public CEO produces quarterly earnings call transcripts and at least one annual shareholder letter that are public, machine-readable and consistent enough in format that pronoun-counting is straightforward NLP. The corpus is built. No new infrastructure required. Truflation, or anyone with a competent data team, could ingest twenty years of S&P 500 earnings calls in a week and produce a back-test before the month was out.

Second, the signal is mechanical. Pronoun frequencies do not require interpretation, sentiment analysis or contestable judgement. They are a count. The ratio is the ratio. Two analysts running the same script produce the same number — a property most "intelligence-driven" indices cannot honestly claim.

Third, the back-test is decisive. If Pennebaker's broader correlation holds in the leadership-population subset — and the academic literature suggests it should — then a long-short portfolio constructed on the I/we ratio, sorted into deciles, should produce a measurable performance differential over a multi-year window. If it doesn't, the negative result is itself valuable, because it tells us where the academic finding stops generalising. Either way, the index produces information that is currently not on any terminal.

What it captures that the standard metrics miss

The deeper point, and the one that makes the index more than a curiosity, is what it indexes about leadership before leadership-quality shows up in numbers. Revenue, margins and ROE describe what has happened. The pronoun ratio describes how the person at the top is positioning themselves with respect to the institution — which is a leading indicator of culture, retention, ego-driven strategic error, succession quality and crisis response. None of those show up cleanly in the standard metrics until they have already damaged the company. Pronoun ratios show up in the next earnings call.

The signal is also, importantly, hard to game. A CEO who decides to "say we more" on a coached basis produces a small, detectable shift in one quarter that does not survive the unscripted Q&A portion of the call, the off-script answers in interviews, or the longer-form annual letter. Pronoun use under cognitive load reverts to baseline within a question or two, which means surface attempts to manage the score are visible to the same NLP that produces the score. The signal is, in a useful way, involuntary. Which is precisely the property a leading indicator needs.

The first cohort, observed

To make this concrete rather than theoretical, here is the starter slate — twenty CEOs from the same broad pool the Deadlift Index draws on, sorted into three bands. The placement below is qualitative, based on publicly observable speech patterns in earnings calls, shareholder letters and on-record interviews. The decimal-precision version is what a systematic NLP build will produce; the version below is what the first issue of the index would look like at first glance, observed by hand.

Collective leaners

CEOs whose public language consistently locates them inside the institution rather than above it. Speech that defaults to "we", "our team", "the firm", "we believe", with first-person-singular reserved for direct ownership of error or specific commitment.

Mixed band

CEOs whose language balances or shifts with context — institutional in scripted moments, more first-person-singular under pressure, in product unveilings, or where the personal brand and the corporate one have grown intertwined.

Singular leaners

CEOs whose public language reliably locates them outside or above the institution. First-person-singular is the default register; the company is the medium of the founder's expression rather than the other way around.

The distribution of the first cohort — twelve to six to two — is itself worth noticing. The first observation any honest version of the index produces is that most S&P 500 CEOs are coached or temperamentally inclined toward institutional voice, and the singular-leaner band is small and concentrated overwhelmingly in founder-CEOs whose personal brand and the company brand have fused. That is not, on its own, a verdict on any of them; founder-mode language can be a tell of vision-driven leadership as much as of ego, and the academic finding is that the signal sharpens when read over many quarters of speech under stress, not from a single observation. But the distribution does suggest that the index, once built systematically, will spend most of its discriminating power among the collective and mixed bands — distinguishing the institutional voice that means it from the institutional voice that has been polished onto a more singular instinct underneath.

Physiological and psychological

The Deadlift Index measures physical capability as a proxy for the executive resilience the spreadsheet cannot see. The 'I-We' Index would measure linguistic posture as a proxy for the executive character the spreadsheet cannot see. The two indices share a single underlying claim — that company performance is downstream of leader behaviour, and that the leader's behaviour leaks through informal signals more reliably than through the carefully managed quantitative ones.

The official metrics are produced by the company. The pronoun count is produced by the CEO without thinking about it. That is what makes it the better signal.

A physiological sibling to a psychological one is what the genre needs to come of age. Two indices in the same family, looking at the same person from different angles — one through the body and one through the language — track the human variable that the financial statements treat as a constant and the markets treat as a residual. The starter slate above is the rough sketch; the systematic build is the next step. But the names are on the page now, in bands, which is what turns a proposal into an index. The first issue is here.