Dan Loeb: The Lost Art of Short Selling, and Why Stock Picking is Back

Why does it matter? Because the same structural force that crushed Google and Amazon bears is now running on Nvidia — and most investors are repeating the mistake.

Dan Loeb built his career exposing internet-era frauds on anonymous chat boards in the 1990s and grew Third Point to nearly $30 billion in AUM by doing one thing consistently: recognizing patterns before the market prices them. Here, he's pointing at something hiding in plain sight.

• Nvidia is undervalued at $3 trillion — long-short pods are structurally forced to short something, and it has become the consensus "safe" candidate, exactly as Google and Amazon once were • Valuation-only shorts are a losing framework; the only safe thesis is structural impairment — balance sheet misrepresentation, post-cycle hangover, forced industry normalization • Tech literacy is no longer a specialization, it's a baseline: "any pool of capital that used to not be correlated is effectively correlated" • The DOJ threatened Trump directly at the end of his first term to kill the Ulbricht commutation — and it worked, for four years

The 'safe short' on Nvidia is the same error that crushed the bears on Google and Amazon

$3 trillion and still cheap. Loeb's call is unambiguous: Nvidia is "absolutely" undervalued on earnings over the next two or three years.

The discount isn't about fundamentals. Long-short pods are structurally required to hold shorts, and Nvidia has become the obvious consensus candidate. "The long-short pods are structured such that they have to be short something. So Nvidia feels like a safe short the way Google was a safe short, Amazon was a safe short." History is clear: both eventually broke through and crushed the bears who treated size as a ceiling.

The market cap anchoring argument gets the same dismissal. Sacks frames the instinct — a multi-trillion dollar company, where does it go from here? Loeb rebuffs it: "I think we'll look back at some point in time and say that was a foolish way to think about Nvidia given its dominant position and its valuation." It's the same cognitive anchor investors used to cap Facebook at $100 billion, back when trillion-dollar companies felt impossible.

When forced sellers drive a discount on a dominant-position company, it resolves. The question is when, not whether.

Valuation-only shorts get destroyed — structural impairment is the only safe foundation

"The lost art of short selling has come back and it's absolutely critical." Loeb says it with conviction — then immediately follows it with a warning that sounds more like a eulogy for undisciplined bears.

Too many short-sellers have been "run over by shorts that have dumb valuations, but they get captured on Reddit." The space sector right now is full of examples — stocks with "no rhyme or reason" to their prices that keep going up anyway. You can be fundamentally correct on the overvaluation and still get margin-called into oblivion waiting for the market to agree.

Third Point's filter: "one thing that we've avoided is kind of a valuation, a solely valuation-based approach." Overvalued isn't a short. Structurally impaired is a short — an industry where the forces suppressing it aren't just "too expensive" but mechanically unsustainable, with a clear catalyst for resolution.

Without that mechanism — the why now — you're making a bet, not building a thesis. In a market where Reddit can keep a broken company aloft for two years, that distinction determines whether you survive long enough to be right.

The homebuilder short worked because the balance sheet contradicted management's own story

The entire homebuilder sector was pretending to be something it wasn't. That gap between narrative and actual balance sheet is exactly where Loeb looks.

NVR runs a genuinely asset-light model, controlling land through option contracts without owning it outright. Every other major builder spent years claiming the same — while sitting on "massive commitments to these land pools which they said were options but they were really very committed in the capital." Asset-light as a marketing story; deeply encumbered on the actual books.

Then layer the macro. Homebuilding was "the last industry that had this post-COVID hangover over of inventory disruptions and pricing that really made no sense." Prices hit unsustainable levels. Inflation squeezed construction costs from the other direction. The financing environment shifted under buyers' feet simultaneously. "Buyers are no longer able to pay those prices at the current financing environment. But there have also gotten squeezed by inflation and costs."

Two forces converging at the same time: a balance sheet the market was pricing on management's narrative rather than the actual commitment structure, and a macro normalization already underway but not yet in earnings. Neither leg alone was sufficient. Together, they were. That's the template.

Tech illiteracy used to be survivable — the opt-out closed and most investors haven't noticed

For decades, some of the best capital allocators in the world operated without a real technology framework. Loeb is direct about when that changed.

"You could be technologically illiterate or just say I don't do it... up until the GFC I think you could be more or less economically illiterate and make a lot of money." Two separate opt-outs, closed at different points in history. The macro opt-out shut in 2008. The tech opt-out has closed somewhere in the last few years — and most investment processes haven't reconfigured around it.

The driver isn't just that tech stocks are large. Correlation has collapsed across the capital structure. "Any pool of capital that used to not be correlated is effectively correlated." A direct lender underwriting a portfolio company needs to know if AI is about to compress its margins. A credit investor in any consumer-facing business needs a view on which revenue models survive disruption. "The tech through line needs to be understood everywhere."

Sacks underscores it: even Blue Owl, built around alternative credit, has to be sophisticated about technology now. Any process that still treats tech as a vertical covered by one specialist is running on a map that expired.

It's a bond and credit pickers market — not a stock pickers market

"Doesn't matter what you do, you have to be really selective. People talk about stock pickers market. This is a bond and credit pickers market."

Loeb drops it and moves on. But it cuts against almost everything financial media emphasizes in high-volatility periods, and it's worth sitting with. The dispersion opportunity — the spread between well-underwritten and poorly underwritten positions — is wider in fixed income right now than in equities.

Third Point's platform architecture makes sense through this lens. The hedge fund has been building out CLOs, private credit, direct lending, credit solutions for distressed situations, and an insurance company capturing the investment-grade end of the capital stack. These aren't diversification plays. They're bets that fixed income selectivity is where the alpha concentrates this cycle.

For strategies built primarily around equity long/short, the uncomfortable question is whether the research effort is aimed at the right market — and whether the return per unit of conviction is actually higher in stocks than in credit spreads right now.

Product moats are a fiction — the only durable edge is management's capacity to adapt

No product lasts forever. Loeb says it plainly, then notes investors have always known this and kept pretending otherwise.

Chamath's "time-bounded value" framing gets a direct endorsement: every business has an expiration window. But Loeb extends the critique backward — the illusion of durable product moats isn't new to the AI era. "I think we deluded ourselves earlier because if you ask people about the moat around IBM... AOL... Yahoo." The confidence always looked right until it didn't. The moat narrative is comfort, not analysis.

What Third Point underwrites instead: "We really look for a management team that we think will be adaptable." The investment thesis is on the people, not the product. Critically, after 30 years, Loeb can't quantify it. Still qualitative. Still pattern recognition built from watching how leaders behave through inflection points, across earnings calls and capital allocation decisions. No rubric captures it.

In an AI era compressing product cycle lifetimes from decades to years, this isn't just investment philosophy — it's the only honest framework. The companies worth holding for ten years aren't the ones with the best current product. They're the ones whose leadership has proven they can find the next three.

The DOJ threatened Trump directly to kill the Ulbricht commutation — and it worked, for four years

On the last day of Trump's 45th term, Ross Ulbricht was supposed to walk free. He didn't.

"The Justice Department, for whatever reason, said, 'If you commute his sentence, we're going to go after you to the president.'" Trump withdrew the commutation. Ulbricht — sentenced to double life plus 40 years for running Silk Road, a crypto-based marketplace where drugs were traded — served four more years.

Loeb had spent years building the coalition that eventually got him out. It started through Intel contacts connecting him to Olaf Carlson-Wii and the broader crypto community. Loeb then brought it to Charlie Kirk, who "really embraced this individual as someone who had been unfairly sentenced" and carried it to the White House through his attorney David Warrington — now White House counsel, who had worked the case for a decade. It was Kirk's "only ask of the president."

The second term delivered. Sentence commuted, then fully pardoned. Ulbricht is now married, expecting a child, living free.

The structural lesson isn't about Silk Road. The pardon process can be neutralized by institutional pressure, even against a sitting president. Getting around it requires a coalition broad and politically durable enough to absorb that opposition — crypto community, conservative movement, White House counsel. Loeb's playbook is a template.

The investors who compound through the next decade are building platforms, not specializations

What Loeb describes, across every thread here, is the end of siloed expertise as a durable edge. You can't wall off tech. You can't rely on product durability. You can't assume equity is where the cycle's alpha lives. And you can't assume institutional processes — whether market structure or the pardon system — behave the way their formal rules suggest.

The investors who win from here won't be the sharpest single-asset stock pickers. They'll be the ones who built platforms capable of harvesting dispersion wherever it surfaces — equity, credit, structured products, insurance — at the moment it concentrates there. Loeb spent 30 years assembling that architecture. Most are still optimizing a single instrument.

---

Topics: hedge funds, short selling, activist investing, Nvidia, homebuilders, credit markets, AI and investing, criminal justice reform, Ross Ulbricht, Third Point, moats, management quality, event-driven investing