Notes — Robert Hagstrom on Pragmatic Truth, Multi-Disciplinary Investing, and the Concentrated Portfolio

Four questions [Adler frame]

Q1 — What is it about? A conversation between William Green and Robert Hagstrom (CIO, Equity Compass; author of The Warren Buffett Way) covering three interconnected subjects: the philosophical framework Bill Miller used to identify Amazon and hold it through a decade of controversy; the multi-disciplinary reading and learning practice that Charlie Munger and Bill Miller share; and the mathematics and psychology of concentrated low-turnover investing.

Q2 — How is it argued? Through biographical anecdote (watching Miller apply Wittgenstein and William James in real time), research citation (Cremers/Petajisto high-active-share papers; Bessembinder on 4% of stocks), and portfolio practice (Equity Compass’s 20-stock, 5-7 year holding period approach). Hagstrom is unusual in having both the theoretical training (Investing: The Last Liberal Art) and the practitioner apprenticeship with Miller; the argument gains force from that combination.

Q3 — Is it true? The Bessembinder data is robust and widely cited. The Cremers/Petajisto finding (high active share + low turnover → excess returns) has held up in replications. The philosophical framework — correspondence vs. pragmatic theory — is a genuine intellectual tradition, not a post-hoc rationalisation. The main vulnerability: Miller’s 2008 collapse (wrong description of government behaviour) proves the framework is not infallible; it is a tool for forming better descriptions, not a guarantee of correctness.

Q4 — What of it? The deepest implication is that the description you choose for a business determines the explanation you construct, and the explanation determines whether you buy, hold, or sell. Most analytical errors are description errors, not calculation errors. Hagstrom also makes a structural observation about the investment industry: the mathematical case for concentration is overwhelming, but prospect theory (losses hurt twice as much as gains please) means most clients, and thus most managers, cannot implement it — creating a durable edge for those who can.

Glossary

Correspondence theory of truth (Wittgenstein, via Miller): the position that our descriptions of the world correspond to some fixed, knowable reality. In investing, the investor who holds a correspondence theory believes there is one correct way to categorise a business (a “value stock” or a “growth stock”) and that departing from that category is error.

Pragmatic theory of truth (William James, via Miller): the position that ideas are tools, judged by whether they work — whether they produce reliable predictions and accurate cash values. The pragmatist describes a business in whatever terms are most predictively accurate, discarding descriptions that don’t work and updating those that do. Value migrates; the pragmatist follows it.

Description/explanation framework (Wittgenstein via Miller): language → description → explanation → action. The description you choose for a phenomenon determines the explanation you reach. If you describe Amazon as a money-losing retailer, your explanation is “avoid.” If you describe it as a negative-working-capital machine (Dell), your explanation is “buy.” The description is the causal upstream.

Active share (Cremers/Petajisto, 2009): the percentage of a portfolio’s holdings that differ from its benchmark. A 100% diversified index fund has 0% active share; a 20-stock concentrated portfolio might have 80–90%. Cremers and Petajisto found that high active share + low turnover was the combination associated with persistent excess returns.

Bessembinder’s 4% (Hendrik Bessembinder, Arizona State): from the Great Depression through 2020, roughly 4% of publicly traded companies account for all net long-run equity returns above Treasury bills. Two-thirds of stocks don’t beat T-bills over their lifetime. The investment problem is therefore a selection problem: identify the 4% before the market has resolved the uncertainty about them.

Competitive advantage period (CAP): Buffett’s most costly recurring error is misjudging the duration of a company’s competitive advantage, not its presence. AI cannot yet assess the CAP; it is the last remaining moat for the human investor.

Inspectional / analytical / syntopical reading (Mortimer Adler, How to Read a Book): a three-level method. Inspectional: skim — read forward, bibliography, last chapter — to determine if the book deserves 8 hours. Analytical: read closely, mark, make the book your own. Syntopical: find all comparable books on the same subject and read them together, building a comparative understanding no single book provides.

Bill Miller’s pragmatic framework

The episode’s core intellectual contribution is a reconstruction of Miller’s applied pragmatism — the version Hagstrom actually watched being used to make investment decisions.

Miller’s starting point is two theories of truth:

• Correspondence: the world works a particular way; successful investors have figured out the categories; stay in your category.
• Pragmatic: what works is what’s true; truth is cash-value; observe what is actually working and figure out why.

The practical upshot: “Value is always in the marketplace. It just migrates.” The correspondence theorist is locked into low P/E, low P/book; value is only in those places. The pragmatist follows value wherever it goes — technology, financials, whatever — because the goal is accurate prediction, not doctrinal consistency.

William James’s blindness essay (1898, “On a Certain Blindness in Human Beings”): James travels to North Carolina, sees a mountain cabin in apparent squalor, and calls it hideous. A local mountaineer corrects him: the cabin is a triumph of human will — family, safety, dignity carved from wilderness. James confesses he was as blind to the ideality of their conditions as they would be to his Cambridge academic life. Miller used this to diagnose Wall Street’s persistent inability to see Amazon correctly: the fund managers criticising the Amazon position had the same constitutive blindness James had about the cabin — a frame so deep they couldn’t see outside it.

The Wittgenstein/Amazon mechanism:

1. Every company can be described in multiple ways.
2. The description chosen determines the explanation reached.
3. Most Wall Street descriptions of Amazon were wrong: first Barnes & Noble, then Walmart.
4. Miller’s description was Dell: negative working capital, customer financing the inventory, 100% return on capital as the consequence.
5. Once the description is correct, the explanation (“buy and hold”) follows automatically.

The 2008 lesson: the correct description of 2008 was not 1992 (savings and loan crisis, where government let equity survive). It was a situation where government committee politics would wipe out all equity in the name of not “lending money to fat cats on Wall Street.” The lesson Hagstrom takes: whenever government is involved in a binary survival-or-destruction decision, the outcome is in the “too hard” pile — no investment thesis can accommodate the political variable.

Multi-disciplinary learning practice

Hagstrom describes the lifestyle that produces multi-disciplinary insight, which is clearly modelled on Miller and Munger but implemented in his own way.

The environment: concentrated, low-turnover portfolio management creates time. 20 stocks, 5-7 year holding periods, no CNBC in the office (Bloomberg for 30 minutes at the open). The “noise” filter is structural, not willpower — the portfolio strategy creates the architecture.

Adler’s reading method in practice: Most books don’t deserve 8 hours. Inspectional reading (forward, bibliography, first and last chapter skim) eliminates most candidates. Of 10 books you think you want to read, 2-3 actually deserve analytical reading. Analytical reading: mark the book, own it, pound ideas in. Syntopical reading: once you’ve identified an important topic, find all related books and read them together.

Bill Miller’s additions: New York Review of Books, London Review of Books, London Literary Supplement — not financial periodicals but intellectual treasure hunts. “I wonder what I can find today.” The payoff is seeing the tangible connection between a non-financial idea and an investment decision (the Dell/Amazon/Wittgenstein chain being the canonical example).

Writing as knowledge ownership: “When you write a book, you own it.” Writing forces full ownership of the material in a way that passive reading does not. The compound effect: ideas written down 25 years ago are still fully accessible because they became part of the writer.

Focus investing: the mathematics and the kryptonite

Bessembinder’s 4% (updated 1990–2020): In the 30-year period, the 50 biggest contributors to total market capitalisation were identified ex-post. Of the 35 US stocks in that group, Hagstrom found that over the most recent 10 years: 17 underperformed in that decade, 17 outperformed. But equally weighted, the portfolio of all 35 crushed the market (up ~600% vs. market ~200%, even excluding Nvidia). The distribution of outcomes: monthly underperformance 50% of time; quarterly underperformance 60%; annual underperformance 63%; 102 separate 20%+ drawdown periods; average peak-to-trough drawdown 41%.

The paradox: the portfolio that produces enormous long-run outperformance has characteristics — periodic underperformance, large drawdowns — that make it nearly impossible for most investors to hold. “Who holds that portfolio?”

Cremers/Petajisto (2009, 2012): academic confirmation of concentrated focus investing: high active share (fewer stocks than benchmark) + low turnover = persistent excess returns. This is Hagstrom’s post-Warren-Buffett-Way validation that the mathematics are not just Buffett’s idiosyncrasy.

Prospect theory as kryptonite: People weight a unit of loss twice as grievous as a unit of gain (Kahneman/Tversky). Rational investors can understand that drawdowns are buying opportunities; almost no one actually buys into a 40% drawdown. This is the structural reason why concentrated investing works despite being the most obvious strategy in finance — the kryptonite is not analytical but neurological.

Bill Ruane’s advice (Sequoia fund founder): “Fire the clients who complain.” Hagstrom didn’t take the advice early enough. Now his strategy: invite clients to put in 10% of intended allocation for 3 years, hands-off. “I don’t want a million and a quarter of your yelling at me when AI doesn’t work this quarter.”

The parallel with private equity: Private equity has no mark-to-market, so clients don’t feel the volatility. Its performance over the last decade is broadly worse than public market indices. Its only remaining selling point is the absence of visible drawdowns. Hagstrom’s proposed solution: run concentrated public portfolios on a private-equity reporting model — report economic returns alongside price returns, so clients can see the business underneath the volatility.

Investment posture and portfolio mechanics

Equity Compass portfolio: 20 stocks ± a few, concentrated in dominant global businesses (Nvidia, Amazon, Microsoft, ASML, Meta, Louis Vuitton). Average holding period 5-6-7 years; 7 of 20 held for 11 years.

Right-sizing bets: Entry position = 2%. Full conviction = 4%. Overbet = 6-8%+. The process is gradual — “I very rarely go from 0 to 4 or 4 to 0.” Continuously seek disconfirming evidence: find the outlier who thinks it goes to zero and the outlier who thinks it goes to the moon; work from both ends towards a probabilistic middle.

Nvidia valuation: Bought fall 2022 knowing AI was coming (Santa Fe Institute GPU/CPU analysis); didn’t know ChatGPT would appear in November 2022. Position grew from 2-3% to 10-12% through appreciation. Had to trim for pension mandate concentration limits. Earnings per share grew faster than share price 2023-2024 — the multiple was actually coming down. Same pattern for Microsoft, Google, Meta, ASML, Amazon. Hagstrom argues these businesses earned their way to trillion-dollar valuations; they are not 1999 Cisco.

AI and the competitive advantage period: Cliff Asness’s claim that AI can do 80% of analyst work is true for factor-based investors (data retrieval, historical modelling, DCF construction). But Buffett’s own stated biggest mistake — misjudging the duration of a competitive advantage — is precisely what AI cannot assess. The question “how long can Nvidia’s advantage last?” requires modelling a financial ecology of competitors, technology trajectories, and strategic responses that exceeds current AI capability. This is the remaining moat.

Bill Miller temperament

9/11: Miller was visiting his alma mater. Learned that AES (bought the day before) had just massively missed earnings — a $50M loss in a day. His response: “Where’s my cash? Let’s double our position.” Rationale: people feel the pain of loss 2-2.5x as keenly as the pleasure of gain; therefore they overreact to bad news; therefore buy. This was applied Kahneman/Tversky, not emotional detachment.

Sherman Kent model: Miller had been an intelligence officer. Sherman Kent (the CIA’s founding analytical theorist) emphasised that the analyst should be multi-disciplinary — bring in scientists, experts from different fields, not just rely on one’s own domain. The analyst’s job is to synthesise, not to specialise. Miller’s natural intellectual style matched this model; Hagstrom speculates it was partly formative.

Baseball pitcher analogy: Miller was a pitcher. Pitchers take public failure as a feature of the job — you lose games, get knocked out of the park, and you go back to the mound. The sport systematically selects for the temperament needed to come back after loss. “If you spent a whole life of pitching… you have to think that helped to craft some sense of not only competition, but perseverance.”

2008 recovery and stoic philosophy: The financial crisis was personally devastating — concentrated in financials, took catastrophic losses. Hagstrom did not know until Green’s Richer, Wiser, Happier how important Stoic philosophy was to Miller’s recovery. The Stoic framework (equanimity in adversity, judgment of what is within vs. outside one’s control) provided an anchor. Miller’s statement: “I’m glad I didn’t curl up like a turtle in its shell. I kept buying.”

Reading as experience

Bill Miller’s claim, as reported by Hagstrom: “All reading is experiential.” What moves people in books becomes an internal experience — a shiver, an aha, a light bulb — that does not come from TV or newspapers. The insight from a book is durable precisely because it is experiential rather than informational.

Hagstrom connects this to writing: the investment adviser who writes down an idea owns it. The commentator who has an opinion without facts (“opinions worth a cup of coffee”) does not own anything. The writer, forced to footnote and defend, owns the intellectual work.

The Sherlock Holmes diagnostic: “It is a capital mistake to theorize before one has data. Insensibly, one begins to twist facts to suit theories instead of theories to suit facts.” Applied to investing: facts come first, description second, explanation third. The investor who starts with the conclusion (this is a value stock / this is a growth stock) and then selects confirming data has the order wrong.