Finding The 1% of Stocks That Matter | Henry Ellenbogen Interview

ELI5/TLDR

Henry Ellenbogen, founder of Durable Capital Partners, discovered that only about 40 stocks out of 4,000 — roughly 1% — compound at 20%+ annually over any rolling ten-year period. He built his entire firm around maximizing the probability of owning those 40. His edge comes down to two things: understanding people (especially “Act 2” entrepreneurs who’ve done it before) and understanding change (knowing which companies will ride a technology wave versus drown in it). The conversation covers how AI is the next great wave, why physical moats still matter, and why being public might actually be better than staying private forever.

The Full Story

The 1% Thesis

Ellenbogen’s origin story starts at T. Rowe Price, where he managed the New Horizons fund — the oldest and most successful small-cap growth fund in the country. He went into the archives and read 50 years of shareholder letters. The finding was stark: only about 20 stocks drove the fund’s entire performance over those five decades.

The anecdote that crystallized everything: one of the prior fund managers had met Sam Walton on Walmart’s IPO roadshow when it had just 50 stores. The fund bought it, then sold it. Had they held, that single Walmart stake would have been worth more than the entire $8 billion fund Ellenbogen was then managing. One sell decision — which felt reasonable on any given day — wiped out the mathematical value of every other good decision combined.

“If there’s 4,000 average public stocks, how many of them truly are great? Over a rolling 10-year period, you have about 40 stocks that compound wealth at 20% a year or go up a little bit over 6x. So about 1% of the stock market are the valedictorians.”

Roughly 80% of those 40 companies start their compounding journey as small caps. That’s the reason for Durable’s focus.

Finding Compounders: Patterns and People

Ellenbogen uses two lenses. The first is pattern recognition — if a company has been a compounder before, the probability of doing it again is higher. His team studies historical case studies obsessively. His partner Anukate teaches a class at Columbia Business School’s value analysis program built around these case studies.

The second lens is people. Ellenbogen is fixated on what he calls “Act 2” entrepreneurs — operators who’ve already built and scaled something successfully and are now doing it again. Workday’s co-founders (who’d built PeopleSoft), Max Levchin (PayPal to Affirm), Luis von Ahn (reCAPTCHA to Duolingo). The advantage of Act 2 teams is they start with total clarity. They know the edge cases. They know how to align investors, org structure, and culture from a clean sheet.

“When you’re an investor in Max and you saw how resilient he was in Slide, and now you understand how he’s got to be resilient to implement AI to lean out his cost and drive his revenue before his competition does — we’ve seen him under stress before.”

Durable almost got named “Act 2 Capital.” That’s how central the idea is.

The Domino’s Lesson: Good to Great

One of the most interesting patterns Ellenbogen describes is what Durable calls “good to great” — existing companies in physical-world businesses that leverage technology to permanently widen their advantage. The best small-cap stock of the 2010s wasn’t a tech company. It was Domino’s Pizza.

Domino’s didn’t grow revenue at 10%+ annually. What it did was invest heavily in its app and direct customer relationships, making convenience a genuine competitive weapon in a market where the middle third — regional chains with mediocre pizza and insufficient scale — simply couldn’t keep up. Technology advantage translated into physical-world distribution advantage, which is much harder to replicate.

The same logic applies to Amazon, Walmart, and Costco. Once Amazon rode the cost deflation curve of e-commerce logistics for 20 years — deflating delivery costs 3-5% annually while competitors were flat or inflating — the gap became insurmountable. Only Walmart and Costco had enough scale and customer trust to eventually get on the same curve. Between them, those three now control 62% of all retail.

AI as Digital Kaizen

Ellenbogen frames AI through a manufacturing analogy. For 40 years, the Danaher Business System (modeled on Kaizen) showed that relentless process improvement in physical-world businesses could compound wealth at 20% annually — even without deep structural moats. AI, he argues, is the equivalent for knowledge work.

“We’ve been able to really lean out product-based businesses and working capital, but in many ways I feel like we’re just getting started on processes that are done by humans.”

He uses Affirm as an example. Max Levchin has hundreds of thousands of merchant contracts requiring legal review, compliance monitoring, credit communication oversight — all human-intensive processes. AI lets Affirm grow revenue without growing headcount. Ellenbogen connected Levchin with Mitch Rales, Danaher’s co-founder, because the parallels are direct: what Kaizen did for factory floors, AI does for corporate processes.

The Duolingo example is even more striking. Luis von Ahn developed chess — Duolingo’s new product — with two people for six months, then four more for three months. Nine months total. Best product the company has ever shipped. Over a million DAUs, growing fast. Previously, the same product would have needed 4-6x the people and 4x the time.

Robotics: The Second Wave

Durable has only recently started documenting its views on robotics — literally within the last month, Ellenbogen says — and he’s candid that these views are “very early and probably deeply wrong.” But the framework is clear:

In certain use cases, robotics is already at cost parity with human labor. And since this is the earliest and worst the technology will ever be, and it’s riding general-purpose AI models (geometric improvement curves, not linear), the Amazon mental model applies. Companies that get on the deflation curve today — 15-20% annual cost reductions — will be two to three years ahead of competitors within five years. And unlike software advantages, physical infrastructure advantages compound: you can’t spin up an Amazon fulfillment center or a Carvana reconditioning center overnight.

Market Structure and the Agency Problem

Ellenbogen estimates 80-90% of institutional flow is driven by firms with one-month or three-month performance horizons, or by quantitative models optimized for those time frames. The result: earnings volatility in Q2 of last year was more extreme than any quarter since the financial crisis — despite fundamentally stable conditions.

His response is deliberate contrarianism. When Colliers International sold off on commercial real estate fears, Durable bought aggressively — because they’d studied Jay Hennick’s capital allocation and organizational culture for 20 years and knew the underlying business was a high-quality asset manager, not a cyclical brokerage. When the 2021 IPO class collapsed 70%+ in 2022, Durable bought more Duolingo — because they’d studied the data on what patterns hold in positive real rate environments.

“The average compounder has a period of time where the stock goes down 50%. And it doesn’t go down 50% only when the market is down 20. They go down 50% when they go through transition.”

The Investment Memo and Accountability Culture

Durable’s memos are structured differently from most firms. For early-stage growth companies, the memo must articulate: if this company does what we think over three years, would we want to buy more at higher prices? If the answer is no, they can’t buy. The thesis can’t be “it gets acquired.”

The firm runs quarterly operating reviews on every position with the full team. But the process Ellenbogen wishes he’d started earlier is the three-year lookback: two slides showing what they underwrote three years ago versus what actually happened. The simplicity is the point. Small quarterly deviations are easy to excuse. Twelve consecutive quarters of drift are impossible to ignore.

Why Going Public Matters

Ellenbogen pushes back on the “stay private forever” thesis with a structural argument. During the zero interest rate era, there were 120 compounders instead of the normal 40 — three times as many — because free money made profitability discipline optional. When rates normalized, companies that hadn’t built the muscle of balancing growth, profitability, and innovation faced an existential reckoning.

The Netflix example is vivid. When Reed Hastings was transitioning from DVD-by-mail to streaming, the stock fell from $280 to $70 (and later to the $50s). Ellenbogen led a PIPE to recapitalize the company — after calling Hastings on a Saturday to walk through a scenario where the transition could burn through all their cash. The public market forced discipline: it made the risks visible, realigned incentives, and ultimately produced a stronger company.

“I actually think to run a company well, you have to be in the and business, not the or business. You have to drive growth, and innovation, and profitability.”

Building Durable (the Firm)

Ellenbogen’s goal for Durable Capital is unusual: make it better the day the founders leave than when they were at their peak. The firm hires young, develops internally, and promotes only people it believes could eventually lead the organization. The team of 12 invests across private and public markets — the same analyst who underwrote Figma at $2 billion private now covers it as a public company worth over $12 billion.

The culture is explicitly “and” — you’re measured on your own investment excellence and on making your colleagues better. 360 reviews require citing specific investments where a colleague’s insight made a difference. Off-sites replaced team-building activities with portfolio lookbacks and industry deep-dives, because that’s what the team actually wants to do.

Key Takeaways

In any rolling 10-year period, roughly 40 out of 4,000 public stocks compound at 20%+ annually. About 80% start as small caps.
Selling Walmart from the New Horizons fund was so costly that the forgone gain exceeded the entire $8 billion fund — one bad sell decision mathematically erased decades of good buy decisions.
During the zero-rate era (2010s), there were ~120 compounders vs. the historical norm of ~40. Free money tripled the odds. That era is over.
“Act 2” entrepreneurs — operators doing their second or third company — have structurally higher odds of compounding because they start with full clarity on edge cases, org design, and capital structure.
Durable’s investment memo test: “If this company does what we think in three years, would we buy more at higher prices?” If not, they don’t invest. The thesis cannot be “it gets acquired.”
AI as digital Kaizen: what Danaher’s business system did for factory floors over 40 years, AI will do for knowledge work. Companies that get on the cost deflation curve early may compound a 15-20% annual advantage.
Duolingo developed its chess product with 6 people in 9 months. Previously: 24-36 people over 3 years. Already over 1M DAUs.
Robotics cost deflation could follow the Amazon playbook — 15-20% annual cost reductions vs. competitors who are flat or inflating. Five years of that gap creates power-law outcomes.
80-90% of institutional flow runs on 1-month or 3-month horizons. This creates earnings volatility beyond what fundamentals justify, which is an opportunity for longer-duration investors.
The average 10-year compounder sees its stock drop 50% at some point — usually during a business transition, not just a market downturn. Surviving that drawdown is the price of admission.
Physical moats (fulfillment centers, reconditioning facilities, franchise networks) are Ellenbogen’s favorite competitive advantage because they can’t be spun up quickly and technology advantages translate into physical distribution advantages.
Durable’s three-year lookback process: two slides, what they underwrote vs. what happened. Twelve quarters of small deviations are easy to excuse individually but impossible to ignore when stacked.

Claude’s Take

This is a dense, high-signal conversation. Ellenbogen is a rare investor who can articulate a coherent philosophy from first principles — the 1% thesis, the biology-as-investing-metaphor, the Act 2 framework — and then show you how it maps to actual portfolio decisions. The Walmart anecdote alone is worth the listen: a single sell decision erasing more value than an entire $8 billion fund is the kind of number that rewires how you think about portfolio construction.

The AI-as-Kaizen framing is the freshest idea here. Most AI investment theses focus on the technology companies themselves — the picks-and-shovels play, the platform wars. Ellenbogen is thinking about what happens to the other 70% of the economy when process improvement that used to require 40 years of Danaher-style discipline can now happen in 5. That’s a genuinely different angle.

Where I’d push back: the Act 2 thesis has survivorship bias baked in. We hear about Levchin and Duffield and von Ahn because they succeeded. The universe of second-time founders who didn’t replicate their first success is large and silent. Ellenbogen acknowledges this obliquely (“in success there’s a lot of good fortune”) but doesn’t wrestle with it much.

The 80-90% institutional flow on short horizons claim is plausible but hard to verify precisely — it’s the kind of number that gets more impressive each time it’s repeated. Still, the directional point is sound: if most capital operates on monthly incentive cycles, anyone with a genuine multi-year horizon has a structural edge, provided they can survive the drawdowns. And Durable’s whole architecture — investor alignment, writing culture, three-year lookbacks — is built to do exactly that.

Score: 8/10. Substantive, well-structured, with several genuinely useful mental models. The conversation earns its length.