Howard Marks Warning: if you invest like this, you're about to lose everything

Why does it matter? Because the best opportunities only look safe in hindsight.

Howard Marks deployed $7 billion in 15 weeks after Lehman collapsed — and freely admits he wasn't confident at the time. That gap between action and certainty is where this conversation lives. Marks covers AI, crisis investing, career randomness, and the anatomy of great partnerships — and the thread running through all of it is the same: certainty is the enemy.

• Fear during a market meltdown is a confirmation signal, not a stop sign — waiting until it disappears means the opportunity is gone
• AI's autonomy makes every prior tech analogy useless, and anyone claiming confidence about its ceiling is overconfident
• The single most dangerous sentence in finance starts with "I'm 100% convinced"
• You cannot raise the capital you need during a crisis — only long before it

Fear is the buy signal — waiting for it to lift means you already missed it

$450 million a week for 15 weeks, deployed into the wreckage of the 2008 financial crisis, with no playbook for what came next.

When Lehman went under, Marks and Bruce Karsh were staring at $11 billion in committed capital and a world talking about total financial meltdown. The logic they ran: if the entire financial system disintegrates, nothing matters anyway. But if it doesn't — and they've sat on their hands — they haven't done their job. So they deployed.

What's striking is Marks' answer when asked whether he felt confident. "No." He was reading the same terrible news as everyone else, feeling the same dread. He just acted anyway.

"A battle hero is not somebody who's unafraid. It's somebody who's afraid but does it anyway." That's the reframe. Fear during a crisis isn't evidence you're wrong — it's evidence the opportunity is real. "If you wait until you have nothing to be afraid about, probably the opportunity has passed." The market clears risk premium by making you feel awful. That feeling is the signal, not the obstacle. If you need calm before you can pull the trigger, you are structurally eliminated from the best returns.

AI is the first technology with autonomy — every historical analogy breaks here

Every prior major innovation — railroads, computers, the internet — was a tool. You told it what to do and it did it faster. AI is the first with autonomy.

Marks puts it plainly: "The idea that you can give it a job and not tell it how to do it and it'll figure it out is really unique." He's spent a career watching new technologies emerge, and he says he never — not through the entire internet boom — felt a technology was beyond prediction. He feels that now. "I have never had that sense before."

His son Andrew, a VC neck-deep in AI companies daily, pushed him to update a cautious memo he'd written in December. Whatever you thought AI could do three years ago is, Marks says, "laughable compared to what it can do today."

On whether AI replaces investors: indexation already "defrocked" the mediocre equity managers who couldn't actually beat the averages. AI will defrock the next layer. The ones who survive are operating where there's no historical pattern to train on — novel situations, first-of-their-kind crises, moments where the data runs out and pure judgment takes over. If your edge is speed or data recall, assume it's gone.

'I'm 100% convinced' is the most dangerous sentence in finance

Mark Twain got there first: "It ain't what you don't know that gets you into trouble. It's what you know for certain that just ain't true."

Marks builds on it with a clean rule: no sentence starting with "I could be wrong, but" has ever gotten anyone into serious trouble. The sentences that destroy portfolios start with "I'm 100% convinced."

The math is brutal. If you're actually in an 80/20 situation and you bet like it's a lock — sizing your position for certainty — the 20% outcome doesn't just hurt. It wipes you out. The overconfidence is the proximate cause of the loss, not the bad outcome itself.

Sam admitted on the spot that he writes his own investment memos like hype pitches — all conviction, all upside. Marks doesn't say certainty is wrong. He says you have to see the other side and size the bet to match your actual conviction level, not your desired one. Epistemic humility isn't a caveat you bolt on at the end. It's a survival mechanism baked into position sizing from the start.

You cannot raise an ark during the flood — Oaktree's $11 billion was built years early

In 2007 — well before the crisis — Oaktree raised an $11 billion distressed debt fund. The prior record they held themselves: $2.5 billion. That gap is the entire lesson.

"The best time to invest is in a crisis. You can't raise money during the crisis because the news is so terrible." The conditions that make assets cheap are the same conditions that send investors running. If you need the capital when everything is on fire, you needed to build it before the smoke.

Marks quotes a line from the movie Spy Game: "When did Noah build the ark? Before the flood." For operators and investors alike, the same logic holds — credit lines, cash reserves, strategic relationships all have to exist before the emergency. By the time you need them, you can't get them. Crisis preparation is not a crisis activity.

Second-level thinking can't be taught — you either see what others miss or you don't

Marks' first book opens with a chapter he wrote spontaneously, without planning: second-level thinking. To outperform, you can't just be smarter or faster at processing consensus information. You have to see something genuinely different — a variant perception — and be right about it.

Can you teach that? His answer, essentially: no. "I can teach you the importance of being a second level thinker. But I can't tell you how to have perceptions that are at odds with the consensus of investors and correct."

In basketball, you can't coach height. Marks thinks insight works the same way — some people have it, some don't, and training can't manufacture it. This is the uncomfortable audit: are you a genuine variant thinker, or are you a fast consensus processor? The latter is the career that indexation already squeezed and AI will finish off.

The world's greatest investors drift too — Howard Marks made unconscious career decisions until age 50

Marks went to Citibank in 1969 because he'd had a good summer there the year before. He moved from equities to bonds because his equity work was unsuccessful and he was told to leave. He moved to California in 1980 for the sunshine and palm trees. He got the call that led him to high-yield bonds because he happened to be at his desk that afternoon.

"If that call came at lunchtime and I had been out at lunch, maybe somebody else would get the call — and they'd be me."

He's blunt about it: up until roughly 1995, when he left to start Oaktree at around age 50, he wasn't making intentional decisions. He was drifting and getting lucky. His advice to young people — figure out what plays to your strengths and makes you happy, don't let society or your parents decide — is advice he admits he didn't take himself for the first three decades of his career.

The odds favor the people who stay in the game

The uncomfortable thread Marks keeps returning to: elite outcomes run on luck more than the biographies suggest. He drifted. The call arrived. He picked up the phone.

What that implies isn't passivity — it's a different kind of preparation. You can't engineer the defining phone call. You can stay present, stay solvent, and stay available when it arrives. Most of the ambitious positioning advice — the career planning, the deliberate skill-stacking — is scaffolding around something that finally just shows up.

The people who get wiped out before the call comes are the ones who bet at 100% conviction on an 80/20 world.

---

Topics: investing, AI, mental models, risk management, partnership, career, Howard Marks, Oaktree, financial crisis, second-level thinking, humility, Warren Buffett