Investing Classics: Reading Howard Marks's MemosFour Memos That Shaped a Generation of Investors

Howard Marks, The Most Important Thing (2003 memo; expanded into a book in 2011).

First-level thinking is linear and effortless: good news → buy. Second-level thinking asks, "Is this information already reflected in the price? Where might the consensus be wrong?" Excess returns can only come from a view that is non-consensus and correct — both conditions are required. Merely being right isn't enough (agreeing with consensus earns no premium); merely being contrarian isn't enough (a wrong non-consensus view loses money). To win the game you must first think differently from others, and be right — which is, by nature, lonely.

The Nifty Fifty. In 1972, U.S. institutions piled into a set of "one-decision, buy-and-never-sell" blue chips: Coca-Cola, Polaroid, Xerox, McDonald's. First-level thinking said, "Great companies — buy at any price." That year Polaroid traded at a P/E around 91, McDonald's around 86, Xerox around 49, versus roughly 19 for the broad S&P. Second-level thinking would ask: however good the business, is there still a margin of safety at this price? In the 1973–74 bear market the S&P fell about 48% from its high, and Polaroid dropped from roughly $140 to about $14 in 1974 (-90%). Great companies, catastrophic prices.

Second-level thinking is not contrarianism for its own sake. A contrarian view without deeper research is just another form of gambling; most of the time the market is right, and reflexively betting against it loses over time. It is also mentally exhausting and "looks wrong" for long stretches — being early feels exactly like being wrong.

• How does my view differ from consensus, and what is that difference based on?
• Can I articulate where the market is wrong, and why?
• Is this good or bad news already fully reflected in the current price?
• If I'm right but the market disagrees for two years, can I hold on?

Howard Marks, "bubble.com" (January 2, 2000) — the first memo to draw a flood of reader responses, written about ten weeks before the Nasdaq peaked.

In bubble.com, Marks made the point: it was true that the internet would change the world, but "the company is remarkable" does not mean "the stock is worth buying." The hallmark of a bubble is that valuation discipline is abandoned entirely — people justify any price with "new economy" and "winner takes all," convinced that "this time is different" (what Templeton called the four most dangerous words in investing). A bubble therefore can't be identified by a precise metric, only by signals of mindset: when "valuation doesn't matter, missing out is the real risk" becomes consensus, the top is usually near.

The Nasdaq peaked at 5,048 on March 10, 2000, then fell to about 1,114 by October 2002 (-78%), with thousands of companies wiped out; Pets.com IPO'd in February 2000 and was liquidated nine months later. The crucial counterpoint: the internet revolution was real, yet the stocks were still a disaster — Amazon fell from about $107 in late 1999 to roughly $6 in 2001 (-94%), even though it later became a giant. The right story paired with the wrong price still produces a loss.

Spotting a bubble is easy; timing it is brutally hard — Marks stresses he doesn't know when a bubble will burst, and one can be trapped a long time while "early but right" (plenty of 1999 short-sellers went broke before the crash). And "this time is different" occasionally really is different, so you can't reflexively short growth just because it's expensive.

• Are people justifying the price with fundamentals, or with a "new paradigm"?
• Has "valuation doesn't matter" become the consensus?
• Am I buying for value, or out of fear of missing out (FOMO)?
• If the story comes fully true but I overpaid, can I still make money?

Howard Marks, "You Bet!" (January 2020), echoing Annie Duke's Thinking in Bets and Nassim Taleb's Fooled by Randomness.

At seventeen, Marks read the line that you can't tell the quality of a decision from its outcome, and it stayed with him for life. Investing is at once a game of skill and a game of chance: you make the best decision you can from what you know, but the result also depends on (a) information you lack and (b) luck. So the best decision-makers aren't those with the most wins, but those with the best process and judgment. Annie Duke calls judging a decision by its outcome "resulting": make money and assume you were right, lose money and reject your whole method — the most common cognitive error investors make.

LTCM (Long-Term Capital Management), 1998. Led by two Nobel laureates (Merton, Scholes), with elaborate models and a seemingly flawless process — yet leverage near 25:1. Russia's August 1998 default, a "tail" event, broke the models; the fund lost over 90% within two months, and in September the Fed organized a roughly $3.6 billion rescue. The reverse is just as instructive: in 1999, many momentum traders got rich on luck, mistook the bull market's gift for skill, and gave it all back after 2000. In the short run, luck can masquerade as skill.

"Focus on process, not outcome" must not become an excuse for bad results — a long, persistent string of bad outcomes usually means the process really is flawed, and can't forever be blamed on luck. Telling the two apart needs a large enough sample and honest review.

• For this gain or loss, how much came from judgment and how much from luck?
• If I had it to do over with the same information, would I decide the same way?
• Do I record my reasons at the time (a decision journal), or only the outcomes?
• Over a long enough span, does my method still hold up?

Howard Marks, "Selling Out" (January 2022), prompted by a conversation with his son Andrew.

Marks notes that investors tend to sell too much, too early. "Sell because it's up" locks in small wins but forgoes the greatest compounding — selling a ten-bagger is a portfolio's largest hidden loss; "sell because it's down" often turns temporary volatility into permanent loss amid panic. The right selling question isn't about the price direction but: at today's price, is this still worth holding relative to other opportunities? Oaktree's principle: as long as attractively priced assets can be bought, stay fully invested, and never move to cash to "avoid risk" — because they don't believe they possess the forecasting ability that market timing requires.

The cost of "missing the best days." Per a J.P. Morgan study, from 2003 to 2022 a $10,000 stake left fully invested in the S&P 500 grew to about $65,000; miss just the 10 best days and you were left with about $30,000 — more than half gone. And those best days tend to sit right next to the worst ones, often just after a panic sell-off. The investor who "sells because it's down" is precisely the one most likely to miss the rebound.

Behavioral finance calls the tendency to "sell winners, hold losers" the disposition effect (Shefrin & Statman, 1985) — it runs exactly counter to compounding.

"Hold for the long term" is not "never sell." The real reasons to sell: ① the original investment thesis is disproven (fundamentals deteriorate, not the share price); ② a clearly better opportunity appears (opportunity cost); ③ a position grows so concentrated it threatens the portfolio's survival. Marks also concedes: trimming an asset that is clearly overvalued at a cycle top is reasonable — not selling ≠ being blind to bubbles.

• Do I want to sell because the price changed, or because the value/thesis changed?
• After selling, do I have a better place for the money (or am I just moving to cash to bet on timing)?
• Is this temporary volatility, or permanent fundamental damage?
• Am I using "taking profit" to cover the cost of forgoing compounding?