Howard Marks on Avoiding Disaster, Risk Posture, and the AI Bubble

Richer, Wiser, Happier

investingvalue-investingriskoaktreehoward-marks

Howard Marks on Avoiding Disaster, Risk Posture, and the AI Bubble

Source: Richer, Wiser, Happier podcast, RWH063 (~late 2024 / early 2025) Guest: Howard Marks — co-founder and co-chairman, Oaktree Capital Management Host: William Green

Occasion: 35th anniversary of Marks’s investment memos; Marks compiled a free digital edition of 45 best memos for the anniversary.

Key ideas

  • If you can avoid the losers, the winners will take care of themselves. Oaktree’s founding motto, derived from a 1990 dinner with a pension fund executive; asymmetry of loss means avoiding blowups creates outsized long-run performance.
  • Fewer losers or more winners — you must choose. The skillful aggressive investor gets more winners; the skillful defensive investor gets fewer losers. Very few have equipment to do both consistently.
  • Risk posture calibration. Every investor should determine their natural risk position on a 0-100 scale (based on age, wealth, income, dependents, intestinal fortitude) and then recalibrate around that posture when markets move to extremes.
  • Taking the temperature. Marks’s five major calls in 50 years were not predictions but observations of investor behaviour — when exuberance is rampant, prices have moved so high that danger is elevated; when depression is universal, cheapness creates opportunity.
  • Idiosyncratic insight cannot survive committees. Great contrarian positions require acting on an insight most people don’t have; any committee can defeat idiosyncratic insight by requiring majority approval.

Background: the 1990 dinner

Marks had dinner in Minneapolis with David Van Benschoten, who ran General Mills’s pension fund for 14 years. Over those 14 years, the equity portfolio was never above the 27th percentile or below the 47th — solidly second quartile, every year. The result: fourth percentile for the 14-year period. The math works because most investors shoot for the stars and occasionally shoot themselves in the foot with a big loss. A big loss takes a long time to recover from; consistent mediocrity compounds into excellence.

This inspired Marks’s first memo (1990): “In equities, if you can avoid the losers and losing years, the winners will take care of themselves.” This became Oaktree’s motto.

Graham and Dodd’s Security Analysis (1940 edition) makes the same point for fixed income: bond investing is a “negative art” — if there are 100 bonds and 90 will pay, the only thing that matters is that you don’t buy any of the 10 that default. You improve performance by what you exclude.

Fewer losers or more winners

From the Marks memo Fewer Losers or More Winners:

You can achieve superior returns by either: (a) having more of the things that go up, or (b) having fewer of the things that go down, or both. Most people can’t do both, because the skills are different — the skillful aggressive player gets more winners; the skillful defensive player gets fewer losers. You must make a conscious choice, and you must know which game you are playing.

Marks also frames this in terms of portfolio posture: Are you maximising the growth of capital or maximising the preservation of capital? You can’t maximise both simultaneously; every portfolio implicitly makes a choice, but most investors never make it explicitly.

Risk posture calibration

From Marks’s memo Calibrating (~8 years prior to the interview):

Think of a speedometer: zero is no risk, 100 is maximum possible risk. Every investor and institution should figure out their default position on this continuum based on:

  • Age and career stage (young, recoverable → higher; near retirement → lower)
  • Wealth relative to income and needs
  • Number of dependants
  • Level of aspiration
  • Proximity to retirement
  • Intestinal fortitude — how much volatility you can genuinely tolerate

Once you know your baseline, you can ask: given current market conditions, should I be above or below my baseline? This is the core question of the Ruminating on Asset Allocation memo (2024).

The insight is not to try to time the market with every change in condition, but to make rare, deliberate recalibrations when the evidence is compelling. Marks estimates he has moved significantly around his baseline only five times in 50 years.

Taking the temperature — five calls in fifty years

Marks’s five major macro calls (based on observing investor behaviour, not predicting outcomes):

  1. January 2000 — Tech bubble: first-day-of-2000 memo on internet.com excess
  2. 2004–2007 — Subprime / mortgage-backed securities excess
  3. 2008 — Deploying $7 billion in distressed debt post-Lehman at $450M/week for 15 weeks
  4. 2012 — (reference in transcript)
  5. 2020 — (reference in transcript)

The method: taking the temperature = assessing investor behaviour, not forecasting events. Buffett’s principle: “The less prudence which others conduct their affairs, the greater the prudence with which we must conduct our own.” When everybody is behaving carelessly and sees no risk, prices have moved to dangerous levels; when everybody is depressed and selling, prices are low enough to be aggressive.

Marks: “I never did it without trepidation. I was never certain, but it felt like the right thing.” He estimates each call was at best 80/20 in his favour, not certain.

If he had tried to make 5,000 calls (one every four days over 50 years), his record would have been 50/50. The rarity of the calls is the discipline.

Idiosyncratic insight vs committees

David Swenson: “Active management strategies demand uninstitutional behaviour from institutions, creating a paradox that few can unravel.”

Marks loathes committees because idiosyncratic insight — the kind that produces great investment outcomes — is by definition what most people don’t see. If most market participants are doing A, how can a majority of a committee vote for B? The insight that most makes you money is precisely the one that is hardest to achieve majority approval for.

The 2008 distressed-debt deployment ($7B in one quarter) would have been impossible to execute through a committee. Marks and Bruce Karsh had pre-raised the capital so they didn’t need to convince LPs in the moment — the structural pre-commitment enabled the idiosyncratic action.

Less efficient markets as source of edge

You can only beat others systematically in markets where hard work and skill can pay off — i.e., where information is not evenly distributed and where most participants are either unable or unwilling to operate.

High-yield bonds (1978): when Marks started, no public pension fund would touch them — they were reputationally and politically unpalatable. “Oh great, it’s an asset class that I can buy that nobody else will buy at any price.” The absence of informed competition means prices can be attractive.

Marks’s metaphor: asking a 1969 PhD with a Cray supercomputer to predict coin tosses at the start of football games. A fair coin cannot be beaten. An efficient market (where everyone has the same information) is the same: no edge. The less efficient market is where hard work and skill create an edge.

The AI bubble parallel

Marks believes the current AI environment most closely parallels the internet bubble of 1998–2000 — not the Nifty 50 (established great companies) or subprime (financial invention, not technological). In both cases: a genuinely transformational technology fired the investment imagination, and “imagination is untraveled” around new things.

Key distinction: in 1999, the vision of how the internet would change the world was fairly clear, and it mostly came true (e-commerce as a major force). Today, the vision of how AI will change the world — specifically, how AI will produce profits (not just productivity) — is less clear.

Buffett’s 2000 annual meeting point: “There’s no doubt that the internet will produce a great increase in productivity. It’s not clear that it’ll have a positive impact on profitability.” The same concern applies to AI. If AI providers compete on price, or if their customers pass savings through to consumers, AI-generated productivity may not accrue to investors.

Marks’s warnings about AI investment mistakes:

  • Don’t assume today’s leaders will be tomorrow’s leaders.
  • Don’t assume cheaper laggards are therefore safer bets (“lottery ticket mentality”).
  • Binary bets (pure-play startups with no revenues) are sink-or-swim; incumbents with existing profitable businesses get moderate benefit if AI succeeds, and stay viable if it doesn’t.

Gold and Bitcoin

Marks’s value-investor position: he cannot invest analytically in any asset that doesn’t produce cash flow. Without cash flow, there’s no intrinsic value to compare against price. Gold can be a store of value, a hedge, a speculation — but you can’t calculate whether it’s cheap or dear.

Data point: gold bought at end of 2010 returned about 7.7% annually through early 2025; S&P 500 over the same period returned about 12.7%. Not a disaster, but not obviously superior to equities.

Charlie Munger

Munger was brilliant, extremely well-read, and had developed a “lattice work of mental models” — a toolkit for pattern recognition. His greatest contribution to Buffett: convincing him to stop buying “okay companies at great prices” (Graham’s cigar-butt method) and instead buy “great companies at okay prices.” That shift is why Berkshire became what it did.

Munger’s other contributions: brutal directness, no patience for committees or bureaucracy, self-deprecating clarity. On his Alibaba purchase late in life: “I just bought it yesterday, so it’s bound to go down 50% immediately.” (It did.)

Balanced life

Marks’s idol Christopher Morley: “There is only one success — to be able to live your life in your own way.” He has played tennis and backgammon, spent time with family, and specifically constructed his role at Oaktree (setting investment philosophy, meeting clients, writing memos) to suit his temperament. “I skip to work in the morning.”