How the great investors actually think: mental models from Buffett, Munger, Dalio and the documented record

Investing has produced more usable, documented thinking than almost any other discipline, for a simple reason: the people who do it well have spent decades writing down how. Warren Buffett has published a shareholder letter every year since 1965. Charlie Munger gave speeches and interviews specifically designed to be re-read. Ray Dalio wrote an entire book called Principles because he wanted his reasoning to survive without him in the room. This is not true of most professions. A brilliant surgeon rarely writes a public memo explaining exactly how they make each incision decision and why.

That difference matters because it means the gap between how great investors think and how everyone else thinks is not mysterious. It is observable, in their own words, at length, across market cycles good and bad. This guide works through the signature mental models behind that record, where they come from, where they hold, and where they break, before showing how differently the best-documented investors actually apply them.

Margin of safety

Benjamin Graham coined the term and Buffett has repeated it in nearly every letter since. The idea is simple to state and hard to practise: only commit capital when the price paid is low enough, relative to a conservative estimate of value, that you would still come out acceptably even if your analysis turns out to be wrong. It is not a prediction that you will be wrong. It is an acknowledgement that you might be, and a refusal to depend on being right.

A worked example: if you estimate a business is worth $100 a share based on its earnings power, margin of safety does not mean buying at $95. It means waiting for something closer to $60 to $70, so that a meaningfully worse-than-expected outcome, higher input costs, a slower recovery, a competitor's better product, still leaves you roughly whole. Buffett's purchase of See's Candies in 1972 is often cited as the transition point where he started paying more for quality than Graham's method allowed, but he still required a price that left room for being wrong about growth.

Where it fails: margin of safety assumes you can estimate value with reasonable confidence in the first place. Applied to businesses with no earnings history, no clear moat, or wildly uncertain futures, the calculation is closer to guesswork wearing a discipline's clothing. It also fails silently in prolonged bull markets, where waiting for a margin of safety means watching opportunities run away for years, which is exactly the period when the discipline is hardest to hold and most necessary.

Circle of competence

Buffett's test is whether you understand a business well enough to have a reasonable view of what it looks like in a decade, and why. He has been explicit that the size of the circle matters far less than knowing precisely where its edge is. His famous refusal to invest in technology companies through most of the 1990s cost him relative performance during the dot-com run, and he said so publicly rather than pretending it was a deliberate hedge.

A worked example: Buffett's large position in Apple, taken decades later, is frequently misread as an exception to his technology aversion. It is better read as confirmation of the rule. By the time Berkshire bought in, Apple behaved less like a speculative technology bet and more like a consumer products company with extraordinary brand loyalty and pricing power, something squarely inside a circle built around understanding consumer moats.

Where it fails: circle of competence can calcify into a reason to avoid anything unfamiliar, including genuine opportunities that simply require learning. The discipline is meant to prevent overconfident forays outside real understanding, not to justify never expanding what you understand. Munger's own reading habits, famously described as sitting and reading much of the day, were explicitly about deliberately widening the circle rather than defending its current borders.

Inversion

Munger borrowed this from the mathematician Carl Jacobi's habit of inverting problems, and made it central to his own method: instead of asking how to succeed, ask relentlessly how you would fail, then avoid those things. He put it more bluntly in his own words: "invert, always invert."

A worked example applied to a hypothetical acquisition: rather than building a case for why the deal will work, list the specific ways it fails, culture clash destroys the target's talent, the acquirer overpays because of deal fever, projected synergies never materialise, key staff leave once earnouts vest. If a plan survives an honest inversion, it deserves more confidence than one that was only ever examined for reasons to proceed.

Where it fails: inversion works on problems with a reasonably bounded set of failure modes. Applied to genuinely novel situations, first-of-a-kind technology, unprecedented market structures, it can produce an illusion of rigour while still missing the failure mode nobody thought to invert because nobody had seen it before.

Long-term orientation

Buffett's stated holding period is, in his words, forever, and Berkshire Hathaway's structure, permanent capital with no forced redemption pressure, is designed specifically to make that credible rather than aspirational. The logic is that compounding rewards patience disproportionately, and that most value destruction in investing comes from unforced trading driven by short-term price movements rather than changes in underlying business value.

A worked example: Berkshire's position in Coca-Cola, built from 1988, has been held through multiple recessions, changing consumer tastes, and long stretches of underperformance relative to the broader market, on the argument that the underlying franchise economics remained intact throughout. The letter commentary from those years repeatedly separates price volatility from business performance as two different things being conflated by the market.

Where it fails: long-term orientation becomes a liability when the underlying business itself is deteriorating rather than merely out of favour. Holding a melting ice cube because you once judged it correctly is not patience; it is the same overconfidence the discipline is meant to guard against, applied to your own prior conclusion instead of a stranger's pitch.

Contrarian conviction

The idea that the best returns come from correct non-consensus positions runs through the letters and interviews of nearly everyone in this category, stated most directly by Buffett as being fearful when others are greedy and greedy when others are fearful. Cathie Wood's public investment theses on disruptive technology apply the same underlying logic to growth rather than value: conviction ahead of consensus, held through the volatility that consensus disagreement produces.

A worked example: Buffett's public buying during the 2008 financial crisis, including preferred stock investments in Goldman Sachs and General Electric on favourable terms unavailable to ordinary investors, was explicitly framed in his own writing as capital deployed because others were forced to be sellers at exactly the wrong moment.

Where it fails: contrarian conviction is indistinguishable from simple stubbornness until the outcome is known, and confirmation bias makes it easy to mistake being early for being right. Michael Burry's public record shows both sides of this: correct and profitable contrarianism on the 2008 housing market, and several subsequent public bets that did not play out the same way.

How they differ: Buffett, Dalio and Wood

Put the three side by side and the incompatibility is instructive rather than a problem to resolve. Buffett underwrites specific businesses through detailed qualitative judgement, holds concentrated positions, and trusts that quality management and durable moats compound over decades. His framework requires believing that some individual judgements, made carefully, are more reliable than any system built to replace them.

Dalio distrusts that premise entirely. Principles is built on the idea that individual judgement, including his own, is unreliable enough that it needs to be checked constantly by systemised rules, radical transparency, and an explicit idea-meritocracy where disagreement is surfaced rather than smoothed over. Bridgewater's "All Weather" approach diversifies across asset classes specifically so that no single judgement, however well-reasoned, can sink the portfolio.

Wood's public theses run in yet another direction: high-conviction, high-volatility bets on technologies she judges to be at an early inflection point, sized aggressively and held through drawdowns that would violate both Buffett's margin-of-safety discipline and Dalio's diversification discipline. All three have produced periods of exceptional performance. All three would tell you the other two are making a category error.

The lesson is not that one is correct. It is that "how to think like an investor" is not a single answer, and anyone selling it as one is skipping the actual disagreement that the documented record shows. What the three share, underneath the incompatible methods, is a written discipline applied consistently rather than intuition applied inconsistently. That consistency, more than any individual model, is what the record actually supports.

---

The Investor category collects these frameworks, and others including Peter Lynch, Howard Marks, Peter Thiel and Nassim Taleb, as downloadable .md files for use with Claude, ChatGPT or any LLM. If you want the full set rather than one framework at a time, The Investor's Stack bundles ten of them, Buffett and Munger included, at the ten-pack rate.