· Howard Marks

Howard Marks: 80 Years of Investing Wisdom in 46 Minutes

Edge comes only from a variant perception you are willing to act on before the evidence is comfortable — and the operating system that makes that possible is capital discipline, probabilistic humility, and a partnership built on shared values plus complementary skills.

investingriskcontrarianpartnershipaicapital-allocationhumility0% confidence

Why this is in the corpus

Howard Marks supplies the corpus with the strongest available articulation of contrarian-plus-patient execution as a repeatable operating system rather than a temperament: pre-building capacity before a crisis, deliberately shrinking funds after wins to buy credibility, and institutionalising "I could be wrong, but" as a risk control. It productively contradicts the high-conviction founder-mode doctrine already dense in the corpus.

Summary for skimmers

Marks explains second-level thinking as the only source of investing edge; how Oaktree raised an $11B distressed fund in 2007 BEFORE the crisis ("build the ark before the flood"); deploying $450M/week for 15 weeks after Lehman with no pattern recognition to lean on; deliberately shrinking follow-on funds after great ones as an anti-asset-gathering credibility trade; the three-part partnership doctrine (shared values, complementary skills, appreciation) behind 39 years with Bruce Karsh; cowboys-vs-chickens as the partnership death spiral; and why AI will defrock investors whose talent is less than they purport but cannot handle events with no history to train on.

Briefing

What survives the editorial filter

This page should feel like a smart colleague already listened for you and left only the operating logic worth keeping. Not everything said in the episode makes it through.

Trust signal

Direct episode extraction

Best used for

Decision-grade retrieval metadata not yet added for this episode.

Hold lightly

No explicit downgrade reason stored yet for this episode.

Principles

Durable claims that survive beyond the speaker's biography — each with explicit limits, transferability judgment, and evidence.

Principle

Certainty, not ignorance, is what ruins people

The damage from being wrong is a function of how certain you claimed to be, because certainty is what sets position size.

Marks: "if you really feel that you're a hundred percent right and you bet like you're a hundred percent right and it turns out it was only 80 20 and the 20 comes up, that's how you get into big trouble." He anchors it in Mark Twain: "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."

Ban "I'm 100% convinced" from your written cases. Replace with an explicit probability, and size to it.

Principle

If you feel no fear, the opportunity is already gone

Comfort is a lagging indicator: by the time an opportunity feels safe, its return has been priced out.

Marks pairs this with the combat analogy: "a battle hero is not somebody who's unafraid. It's somebody who's afraid, but does it anyway." The implication is that fear should be treated as confirmation the opportunity is still live, not as a veto.

Treat residual fear as evidence the window is still open. Require yourself to act while some of it remains.

Principle

Randomness makes short-run track records nearly uninformative

A good year is a draw, not a signal — evaluate process over a horizon long enough for luck to wash out.

Marks: "you see a published record, A guy had a great return that year, is he a great investor? Did he get lucky that year?" He applies the same standard reflexively to himself, attributing his entry into high yield bonds to a phone call he happened to be at his desk to answer.

When evaluating an operator or a fund, ask what the process was and how many independent decisions the record contains — not what the return was.

Principle

Decide by asking which error you can survive, not which outcome is likelier

Prune the branches where your decision does not matter; decide on what is left.

Faced with the Lehman collapse and no way to forecast systemic outcome, Oaktree noted that in the meltdown scenario their deployment decision was irrelevant, so the only live scenario was survival — in which failing to deploy was a dereliction. Marks pairs it with a quantitative floor: they were buying debt of LBO'd companies where "if they ended up being worth a fifth or a fourth of what they had paid, we would still be okay."

For your hardest call, list the scenarios in which the decision is irrelevant and delete them. Decide on the survivors only.

Principle

Appreciation is the third partnership element, and it is the one people omit

Complementarity decays into resentment unless it is continuously re-valued out loud.

Marks names mutual respect as "the bedrock of our relationship" and observes the same structure in Buffett and Munger — a relationship "always suffused with humor" in which Buffett publicly cast Munger as the big brother. The generosity about the other's role is treated as maintenance work, not sentiment.

Name, out loud and specifically, the work your partner does that you would hate doing. Do it often enough that they believe it.

Principle

A variant perception is worthless unless you bet on it

Contrarian insight, contrarian action, and being correct are three separate hurdles, and all three must clear.

Most people who claim a contrarian view never size a position behind it, which makes the view costless and therefore untested. Marks's own record consists of five macro calls over 26 years — a very low rate of acting, which is itself the discipline.

Count how many of your contrarian views you have actually staked capital or career on. That number, not the number of views, is your edge.

Principle

The market's real function is to reject bad ideas; when it stops, the crisis is already being financed

Crises are manufactured during the easy-money period, when the market stops refusing bad ideas.

Marks describes the market's main job: "people come in and say, I want money for this, this and this. And the market's job is to say, no, that doesn't make any sense. That's a stupid idea." Oaktree used exactly this observation — the market not acting as disciplinarian — as the argument that convinced LPs to fund the 2007 distressed vehicle.

Track how easy it is for weak proposals in your market to get funded. Rising ease is your cycle clock, not sentiment.

Principle

Build the ark before the flood — capacity must be raised before the crisis it exploits

The window to raise the resources for a crisis closes the instant the crisis begins.

Oaktree raised an $11B distressed debt fund in 2007-08 — more than four times the largest distressed fund in history to that point (their own $2.5B 2002 fund) — and held it on the shelf specifically for deployment after the break. Marks quotes the Spy Game line: "when did Noah build the ark before the flood?"

Identify the resource you would need most in your worst plausible quarter, and acquire it now, while the case for it is unpopular and cheap.

Principle

Acting with trepidation is the correct state; its absence is the warning

Calm confidence during a big bet signals that you have stopped thinking probabilistically, not that you are right.

Marks notes that during the Lehman deployment — arguably his most successful call — "we were absolutely not confident." He reads the same newspapers as everyone else, sees the same terrible news, and overcomes it without becoming immune to it.

Before a major commitment, check whether you feel any unease. If none, find the point where you replaced a distribution with a certainty.

Principle

Shared values and complementary skills are the two load-bearing conditions of a durable partnership

Values keep partners from fighting; complementary skills keep partners from becoming unnecessary to each other.

Marks and Bruce Karsh have been partners 39 years with intellectual disagreements but, by his account, never a fight. He attributes durability partly to neither being a financial maximizer — removing money, the most common fight topic, from the surface area. The division is concrete: Marks goes on the road and does podcasts; Karsh stays back and manages the money.

Test a prospective partner on two axes separately: would we resolve a hard call the same way, and can they do something I genuinely cannot?

Principle

Edge requires a variant perception, not better information

You cannot outperform a consensus you agree with; superiority requires seeing something different and being right about it.

Marks frames this as the first chapter of The Most Important Thing — written as a sample chapter for Columbia and unplanned. It is not a claim about working harder; it is a structural claim about where excess return can come from at all. The practical consequence is that every position should be interrogated for what specifically it assumes that the market does not.

For any bet, write down the specific belief you hold that the consensus does not. If you cannot name it, you have no edge.

Principle

Speaking against your own interest is what buys credibility you can spend later

Turning down money when the case is weak is what makes people believe you when the case is strong.

Marks describes the payoff explicitly: "people tend to say, when Howard and Bruce say there's a great opportunity, they're not just trying to raise money, they really believe it and they're, and they tend to be right." That belief is precisely what made an $11B pre-crisis raise possible.

Build a documented history of saying "not now" so that your eventual "now" is priced as information rather than salesmanship.

Frameworks

Reusable systems and operating models — including when they help and when they break.

Framework

Data, analogies, supposition — classifying what a decision actually rests on

Classify your inputs as data, analogy, or supposition — supposition-only decisions require a different method entirely.

Marks credits the taxonomy to a Harvard epidemiologist speaking during the pandemic and applies it retroactively to Lehman, where Oaktree had neither data nor precedent. The framework also explains his AI scepticism: pattern-recognition systems trained on history are strongest exactly where humans are, and blind where only supposition exists.

Label each major decision by its substrate. If it is supposition-only, stop forecasting and start reasoning about asymmetry and survivability.

Framework

The three-element partnership diagnostic

Audit partnerships on three independent axes — shared values, non-overlapping skills, and expressed appreciation — each with its own failure mode.

Marks names the redundancy failure precisely: eventually the partner who can do everything says "you're overpaid. I don't need you." He and Karsh institutionalised complementarity from the start — Karsh brought the distressed debt idea in 1987 from a legal background, Marks came from high yield bonds in 1978, and the external/internal split has held for 39 years.

Run the three-element audit on your current partnership annually. Weakness on any one axis is sufficient to end it.

Framework

The four-axis variant perception test

A usable contrarian thesis must name which of quality, growth, earning power, or multiple the consensus has mispriced.

Marks presents this as the operating content of second-level thinking rather than a separate tool. The framework is symmetric — you can be variant by believing consensus overstates or understates any of the four — and the discipline is that you must pick one and defend it, not gesture at general mispricing.

Before taking a differentiated position, write one sentence naming which of the four the market has wrong and why.

Signals

What appears to be shifting, for whom it matters, and what happens if you ignore it.

Signal

"I'm 100% convinced" as a live risk indicator in your own team

Absolute-certainty phrasing in memos is an early, readable indicator of a sizing error about to be made.

Marks contrasts the safe forms — "I could be wrong, but" and "I don't know, but," which he says never got anyone into trouble — against the dangerous form. The transcript supplies a live specimen: a memo urging "bet everything you have, this is it. Mortgage the house."

Flag any memo containing absolute-certainty phrasing for a sizing review before the position is taken.

Signal

The market stops saying no

When weak proposals start clearing the funding bar, you are already inside the setup for the next crisis.

Oaktree used this as their central evidentiary argument to LPs in 2007. It was not a forecast of timing but a statement about observable underwriting behaviour, which is why it was persuasive enough to raise $11B ahead of the event.

Track the quality of the marginal deal getting done in your market. When you can no longer explain why it was funded, start pre-positioning.

Opportunities

Only included where there is a buyer, a real wedge, and a plausible revenue path — not vague idea theater.

Opportunity

AI will defrock the pretenders — position as the demonstrably real thing

AI functions as a verification layer — it destroys unverifiable skill claims and sharpens the signal for genuine ones.

Marks draws the direct precedent: indexation "put a lot of people out of the equity business because, you know, it, it disclosed that they couldn't do what they claimed to do." Most active equity managers underperformed the average, and cheap index products made that legible. He frames AI as the same event for the next cohort.

Identify which part of your value is unverifiable today, and assume it becomes measurable. Move your positioning onto the part that survives measurement.

Opportunity

Judgment in domains with no history to train on

The durable human niche in the AI era is the class of decisions where there is no history to extrapolate from.

Marks grounds this in Lehman, where Oaktree had "no data and no prior experience, we only had supposition," and asks directly whether AI could have engaged in that kind of thinking. He also names a second residual niche — character judgment, "the hair on your neck goes up" — noting that Oaktree helped clients most "by not investing with bad people," and observing that if AI has no neck hair, experienced judgment retains a role.

Audit your work for how much depends on precedent. Shift weight toward no-precedent judgment and character assessment.

Lessons still worth keeping

Useful takeaways that did not fully clear the bar for durable principle status.

Lesson

Marks spent 25 years making unconscious decisions and says so plainly

A canonical career can be built on 25 years of unintentional decisions; the discipline can be installed late.

His examples are unglamorous: Citibank because he had a good summer there; the bond department because he was told to leave equity research after failing; California in 1980 for "sunshine palm trees." The high yield franchise came from a call he happened to be at his desk for — "if that call came at lunchtime And I had been out at lunch, maybe somebody else would get the call and and they'd be me."

Stop treating your unstrategic early years as disqualifying. Start applying intention now; the compounding window is longer than survivorship stories suggest.

Lesson

Munger's single greatest contribution was talking Buffett out of his own strategy

The largest return from a partnership can be a single successful argument against the founder's own strategy.

Marks describes Buffett's original cigar-butt method — picking up a discarded cigar with "three puffs left" — and Munger's reversal of it, which reframed Berkshire from bargain scavenging to quality compounding. He also notes the ongoing function: "I think Warren used Charlie as a sounding board, a logic checker."

Give one person standing to challenge your core strategy, and check whether they have ever actually used it. If not, they do not have it.

Lesson

When both options are acceptable, hand the decision to the person who has to live with it

Delegate every decision whose worst outcome is acceptable, because the reps are worth more than the marginal quality.

Marks and his wife had a preference between two Los Angeles upper schools but concluded "we could be wrong. Hard choice could be the wrong choice," and let their daughter decide. He explicitly values the possibility of a wrong choice: "maybe they get a experience with making incorrect choices, which is very important."

Screen delegation by downside, not by importance. Narrow downside spread means hand it over, including the right to get it wrong.

Lesson

A rare high-leverage offer must be acted on immediately, not scheduled

Offers contingent on another person's attention are perishable; execute against them on their clock, not yours.

Buffett told Marks he read the memos and offered a blurb if Marks wrote a book. Marks abandoned his retirement plan and wrote The Most Important Thing, which became his defining public artifact. The relationship itself started the same way — Buffett's "if you ever find yourself in Omaha, let me know" was answered with an immediate letter proposing lunch that week.

When someone with rare leverage extends a soft offer, convert it to a concrete date within the week. Soft offers do not survive scheduling.

The Plays

Try these this week

Verb-first executable actions — each one tied to a stated outcome in the episode.

Sell the countercyclical vehicle as a hedge on the rest of the portfolio

Outcome: Pitch the crisis fund as protection for the buyer's existing prosperity-tilted book, not as a forecast they must accept.

Context: Marks pairs this with two supporting arguments: a 20-year reservoir of goodwill and prior crisis performance in 1990-91 and 2001-02, and specific named flaws in the environment — chiefly the market failing to act as a disciplinarian. Together these raised $11B.

“so many of your investments are set up for prosperity. This is a good way to hedge it by making an investment that will do particularly well if, if the stuff hits the fan.”
— Howard Marks

During the benign part of the cycle, before the dislocation per

Stop or pivot when

→Only run this where you have documented performance in a prior crisis
→Only where you can name concrete structural flaws, not just a feeling

Scripts

Before you start

· Prior crisis track record
· Long relationships and accumulated goodwill
· A named execution partner with independent credibility

Hear the panicking operator out in full, then send them back to their desk

Outcome: Extract the full substance of a subordinate's panic, absorb it, and then decline to change the mandate.

Context: In 1998 — LTCM, the Russian ruble crisis, the Southeast Asian panic — a skilled young Oaktree portfolio manager told Marks "I think this is it. I think we're melting down. It's all over." Marks solicited the full case, acknowledged it, and returned him to work. He frames it through the combat analogy: the hero is afraid and acts anyway.

“And I said, well, tell me your concerns. And he laid out his concerns and I said, okay, I I understand it now go back to your desk and do your job.”
— Howard Marks

A single conversation per

Stop or pivot when

→Use where the person is capable and the panic is emotional rather than evidentiary
→If the concern surfaces genuinely new information, change the view rather than the person

Scripts

Before you start

· Established authority and trust
· A firm-level view that has already been stress-tested

Shrink the follow-on fund after a great one

Outcome: Great results mean the opportunity has narrowed; size the next fund down and bank the credibility.

Context: Marks describes running this pattern for 20 years before 2007. The accumulated effect is the reason the $11B raise worked: LPs concluded that when Oaktree said there was a great opportunity, it was not a fundraising posture.

“Now most people in the investment business, if they have a fund that does great, the next fund is binger 'cause they can sell on the back of those results. But we make it smaller because we think those results mean that things have appreciated and are not so attractive.”
— Howard Marks

Each fund cycle; credibility compounds over roughly 20 years per

Stop or pivot when

→If realised returns came from multiple expansion rather than new opportunity, shrink
→Only upsize when you can name the structural flaw creating the opportunity

Scripts

Before you start

· Willingness to forgo management fee revenue
· LP base that rewards discipline over AUM growth

Pre-raise crisis capacity and hold it on the shelf

Outcome: Raise the crisis vehicle two years early, accept the drag, and deploy it after the break.

Context: Oaktree's prior high-water mark for a distressed fund was $2.5B in 2002; the 2007-08 vehicle was $11B, sized to a thesis about coming distress rather than to current opportunity. The fund sat undeployed until Lehman.

“in oh seven, eight, we raised 11 billion for distressed debt fund because we thought that that there was a lot of distress coming and we had it on the shelf. It wasn't, it was for deployment when the stuff hit the fan”
— Howard Marks

Raise 12-24 months ahead of the expected dislocation per

Stop or pivot when

→Only raise when you can name specific structural flaws in the environment
→Deploy only when purchase prices survive a 75-80% impairment assumption

Scripts

Before you start

· Prior crisis track record
· Reservoir of LP goodwill
· Credibility from having previously declined to raise

Deploy through a panic on a fixed weekly cadence

Outcome: Convert a crisis deployment into a mechanical weekly rate so the panic cannot veto each individual decision.

Context: Oaktree put $7B to work in a single quarter after Lehman, at roughly $450M a week for 15 straight weeks, while Marks says "we were absolutely not confident." The cadence is what let a not-confident firm still act at scale.

“Bruce, who runs those funds, invested an average of $450 million a week for 15 weeks, 7 billion in a quarter on that.”
— Howard Marks

One quarter to deploy a multi-billion vehicle per

Stop or pivot when

→Buy only where you break even if the asset is worth a quarter to a fifth of the prior LBO price
→Suspend if prices recover above the floor

Scripts

Before you start

· Pre-raised committed capital
· A partner who can execute without needing daily conviction resets

Open every investment case with "I could be wrong, but"

Outcome: Institutionalise an explicit uncertainty clause at the top of every written investment case.

Context: Marks reports doing this reflexively across five macro calls in 26 years. Host Shaan Puri says on air that he will now start his own investor memos with "I could be wrong, but," describing his default style as boastful, exuberant, and certainty-laden — a direct read on how the habit transfers.

“we we're the kind of people who always say, I could be wrong or it could work in a, in a way that's never been seen before.”
— Howard Marks

Immediate; habit forms over a few memo cycles per

Stop or pivot when

→If you cannot name a failure branch, the memo is not ready
→If stated probability exceeds ~90%, re-examine what you assumed away

Scripts

Before you start

· A culture that does not penalise stated uncertainty
· Written rather than verbal decision records

Decision Moments

Actual decisions, real outcomes

Specific decisions narrated in the episode with their outcomes and transferable lessons.

In 2007-08, with credit markets still functioning and sentiment benign, Oaktree believed significant distress was coming but had no way to prove timing. The largest distressed debt fund in history to that point was their own 2002 vehicle at $2.5 billion.

Did: Raised an $11 billion distressed debt fund ahead of the crisis and deliberately held it undeployed on the shelf. The pitch rested on four legs: a 20-year reservoir of goodwill; demonstrated crisis performance in 1990-91 and 2001-02; a reframe of the vehicle as a hedge on LPs' prosperity-tilted portfolios rather than a macro forecast; and specific named flaws in the environment, chiefly that the market had stopped acting as a disciplinarian.Outcome: The fund was in place and fully committed when Lehman failed, enabling deployment at the bottom. Marks notes the capacity could not have been assembled after the fact: "You can't raise money during the crisis because the news is so terrible."

Capacity acquisition and capacity deployment sit in opposite halves of the cycle. Build the ark before the flood — and sell the ark as insurance on what the buyer already owns, not as a forecast they must accept.

Part of an emerging decision pattern across multiple episodes

September 15, 2008. Lehman Brothers has failed. Commentators are describing the end of the financial system — that all financial institutions will melt down and everything to do with money will atomize. Oaktree holds $11 billion in committed, undeployed capital and must decide whether to invest it. Marks: "there's no pattern recognition for the end of the world."

Did: Pruned the outcome tree to the branch where the decision mattered: if the financial world melts down and we invest, it does not matter; if we do not invest and it does not melt down, we failed at our job. Backstopped this with a quantitative floor — buying debt of private-equity-owned companies where they broke even if the businesses were worth a fifth to a fourth of what had been paid. Then deployed on a mechanical cadence, averaging $450 million a week for 15 consecutive weeks, roughly $7 billion in a quarter.Outcome: One of the defining deployments of the crisis, executed while Marks says "we were absolutely not confident." He reports having no data and no prior experience — only supposition.

When only supposition is available, stop forecasting. Delete the scenarios in which the decision is irrelevant, set a quantitative floor that survives severe impairment, and convert execution into a fixed rate so daily panic cannot veto each individual trade.

Part of an emerging decision pattern across multiple episodes

Repeatedly across the 20 years before 2007, Oaktree correctly called an opportunity, raised a large fund, deployed it and made a great deal of money. At that moment the firm could have raised substantially more on the strength of the results — the industry default.

Did: Deliberately made the next fund smaller, reasoning that strong realised results meant assets had appreciated and the forward opportunity set had narrowed. Accepted the forgone management fee revenue and explained the reasoning to LPs rather than selling on the back of the track record.Outcome: Over 20 years this compounded into the specific credibility that made the $11 billion pre-crisis raise possible. Marks: "people tend to say, when Howard and Bruce say there's a great opportunity, they're not just trying to raise money, they really believe it."

A great result is evidence the opportunity has shrunk. Systematically declining money you could raise is the mechanism that converts discipline into the ability to raise an outsized amount when it actually matters.

Part of an emerging decision pattern across multiple episodes

1998: Long-Term Capital Management collapses, the Russian ruble crisis is underway, and Southeast Asia is in panic. A skilled young Oaktree portfolio manager comes to Marks and says: "I think this is it. I think we're melting down. It's all over."

Did: Did not reassure and did not argue. Asked him to lay out his concerns in full, listened to the complete case, acknowledged understanding it, and then sent him back to his desk to do his job with the mandate unchanged.Outcome: Oaktree held its positions through the 1998 dislocation. Marks frames the norm through the combat analogy: "a battle hero is not somebody who's unafraid. It's somebody who's afraid, but does it anyway."

Panic in a capable person bundles real information with emotion. Extract the information fully so it can be absorbed, dissipate the emotion by hearing it rather than rebutting it, then hold the mandate. Establishing that fear is expected and does not change the job is what carries a firm through a drawdown.

Part of an emerging decision pattern across multiple episodes

Tensions surfaced

Contradictions and trade-offs the episode raises — judgment calls a thoughtful operator has to navigate.

Tension

Marks built a teaching franchise on a skill he says cannot be taught

The frame for an elite skill is teachable even when the generative capacity behind it is not — which turns the problem from training into selection.

Marks: "I can make someone better, but I don't think I can make 'em great," and "in basketball there's a saying you can't coach height. And I think there's something called insight." He also allows a partial escape hatch — "There are just some people who have a better understanding of the probability distribution that defines future events" — which is a selection criterion, not a curriculum.

Separate your development budget from your selection budget. Teach the frame broadly; hire for the insight you cannot install.

Tension

You must act on a variant perception you cannot be confident in

Large bets require conviction to execute and doubt to survive, and the two must be assigned to different parts of the decision.

Marks made the biggest deployment of his career while explicitly not confident, and describes all five of his major macro calls as made "with some doubt." The resolution is procedural: the asymmetry argument and the quantitative break-even floor carried the action, while the doubt constrained sizing and kept the position revisable.

Split the question: is the asymmetry good enough to act at all, and separately, how large can this be if I am wrong?

Tension

Institutionalised self-doubt versus the conviction operating mode of high-agency founders

Institutionalised hedging is correct where you cannot influence the outcome and costly where your conviction is itself a causal input.

The host, an operator, names his own default as boastful and exuberant and says Marks has "infected me with a little bit of your humility" — then immediately recalls a partner memo saying "bet everything you have, this is it. Mortgage the house." That is the founder mode Marks's doctrine directly contradicts. The corpus contains a substantial body of high-conviction founder doctrine; this insight is the counterweight and the boundary condition on both.

Apply hedged language where you are a price-taker on the outcome, and reserve stated conviction for domains where belief is causally load-bearing.

Corpus connection

Where this episode fits for retrieval

What kinds of decisions this briefing is best pulled into.

Primary decisions

• strategic-bet
• capital-allocation
• partner
• timing