Uncertainty & Prediction

Concept Analysis

Definition & Origins

The foundational acknowledgment that the macro future is unknowable — that economic forecasting is unreliable — and that the appropriate response is preparation over prediction, building portfolios that survive a range of outcomes rather than optimizing for a single forecast.

Marks' engagement with uncertainty begins with his credit analyst training: the question of whether a borrower will repay is irreducibly probabilistic, not deterministic. This is different from equity analysis, where the standard framework often treats the future as knowable. Credit analysis forces honest engagement with what cannot be known.

The 'Nobody Knows' memo (2001) was written immediately after September 11 — a quintessential black swan event that no forecaster had predicted. The impossible-to-predict character of that event became the template for Marks' general argument: if a macro outcome of that magnitude is unforeseeable, what does that imply about the reliability of any macro forecast?

But the earliest full-dress version of the argument predates that series by nearly a decade. In 1993 Marks wrote "The Value of Predictions, or Where'd All This Rain Come From?" — the subtitle a jab at the California drought of the early 1990s, which tree-ring analysts had declared might be the new norm just as it was ending. The memo frames forecasting as an economic activity with an expected value: the value of a correct forecast multiplied by the probability of being correct. Most forecasters, he argued, fail on the second term. Most predictions are extrapolations of the recent past; extrapolation is usually right; and what everyone expects is already in prices. The memo already carried the Galbraith line Marks would still be citing three decades later — the world divides into forecasters who don't know and forecasters who don't know they don't know.

Core Ideas

The future is genuinely unknowable, not merely unknown. Marks distinguishes between 'unknown' (information that exists and could be discovered) and 'unknowable' (outcomes that have genuine randomness or are causally undetermined). Most macro events are in the second category. The Federal Reserve's next decision is unknown but exists in someone's mind; the outcome of a geopolitical confrontation is genuinely unknowable because it depends on many actors' decisions made in response to each other's actions.

Forecasting has no demonstrated value in aggregate. Despite the enormous investment of time, money, and talent in macroeconomic forecasting, there is no convincing evidence that forecasters, as a group, outperform simple extrapolation over multi-period horizons. The illusion of forecasting skill is maintained by selective recall of hits, adjustment after the fact, and the persistence of people who were right once.

Macro-agnosticism is not passivity — it is a positive repositioning toward what is knowable. Marks is not arguing that investors should be passive or indifferent to the macro environment. He is arguing that capital allocation should be based on what is genuinely analyzable: security-level fundamentals, credit quality, valuation relative to intrinsic value, cycle position in terms of sentiment and supply/demand — not on macro forecasts.

You can prepare for a range of outcomes without predicting any one of them. Portfolio construction under uncertainty requires asking: what scenarios are plausible, how would this portfolio perform under each, and is the aggregate risk/return profile acceptable? This is different from optimizing for a single forecast. A portfolio that survives the 1-in-10 bad scenario while participating in the 7-in-10 good scenarios is better designed than one that maximizes the good scenario and fails catastrophically in the bad one.

The appropriate response to uncertainty is humility about one's own process. Uncertainty doesn't just apply to the external world — it applies to the investor's own analysis. Even within the domain of credit analysis where Marks operates, forecasts of recovery rates, restructuring outcomes, and business recovery are uncertain. The correct response is wider margins of safety, more conservative assumptions, and more explicit scenario analysis.

Being right along with the consensus adds nothing; being right against it is nearly impossible. This is the Catch-22 at the heart of the 1993 memo. Most forecasts are extrapolations, most extrapolations are roughly correct, and a roughly-correct consensus view is already reflected in security prices — so a correct consensus forecast earns only the normal return. Superior performance requires a correct non-consensus forecast, but non-consensus forecasts are usually wrong, because the consensus is usually right. Extreme predictions are rarely right, but they are the ones that make big money — and they are hard to make, hard to make correctly, and hard to act on precisely because they sit so far from conventional wisdom that no one can bring themselves to believe them.

A forecast is never enough; you also need a view on its reliability. In the 2020 memo "Uncertainty," Marks insists that an opinion about the future must be accompanied by an honest estimate of the probability that the opinion is correct. Some events can be predicted with substantial confidence — will an investment grade bond pay the interest it promises? Some are uncertain — will a given company still lead its industry in ten years? And some are entirely unpredictable — will the stock market go up or down next month? Chain logic makes the point concrete: a portfolio built on the reasoning that the economy will do A, so rates will do B, so the market will do C, so sector D will lead, depends on every link holding. Even at a generous two-thirds probability per link, the full chain is right only about 13% of the time. Anyone who is sure about what is going to happen in the world, the economy, or the markets is probably deceiving himself.

Practical Application

Nobody Knows (2001): The September 11 memo argued that in a moment of genuine macro uncertainty, the appropriate portfolio posture is not to bet on an outcome but to ask what prices imply about outcomes and whether those implications are more or less pessimistic than warranted. After September 11, prices implied extreme pessimism — which meant expected return was high for someone willing to provide liquidity.

Nobody Knows II (2020): The COVID version made the same argument in a faster-moving scenario: prediction about COVID's trajectory was impossible in February 2020. Marks did not predict recovery. He argued that at the depth of the market fall, pessimism had been priced in aggressively enough that the risk/return ratio was attractive even without knowing the trajectory.

The Sea Change memos (2022–2025): The regime change of 2022 is itself an argument from uncertainty: Marks does not claim to know where interest rates will go. He claims that the world has changed enough that the assumption of 'rates return to zero' — which many portfolios implicitly rely on — is no longer tenable. The appropriate response is not to forecast rates but to reduce the portfolio's dependence on any single rate environment.

Nobody Knows Yet Again (2025): The tariff shock of April 2025 produced the fourth iteration of the argument. Marks' observation is that there are no experts on the subject at hand: there have been no large-scale trade wars in the modern era, so the theories economists bring to bear are untested, and no economist's conclusion can be followed with confidence. Under such circumstances forecasts are even less likely to prove correct than usual — yet investors cannot wait for certainty, because deciding not to act is not the opposite of acting but an act in itself, to be scrutinized as critically as any trade. The discipline is unchanged from 2008 and 2020: reason out what is most logical, act without pretending to confidence, and keep the portfolio survivable under outcomes other than the one you expect.

Common Misconceptions

Misconception 1: If you can't forecast, you can't invest This is the error of treating uncertainty as a barrier rather than a permanent condition. Every investor operates under uncertainty. The question is whether you acknowledge it explicitly (and build portfolios accordingly) or pretend it doesn't exist (and build portfolios that are catastrophically fragile to the outcomes you haven't modeled).

Misconception 2: Confidence is a virtue in investing Social and institutional contexts reward confident forecasts. 'I don't know' is interpreted as weakness. But epistemic humility is not the same as analytical weakness — it is an accurate description of the state of knowledge about the macro future. The investors who communicate false confidence about macro outcomes are not smarter; they are more willing to mislead.

Misconception 3: Forecasters can be identified by their track records They could be — if anyone published the records. Marks has noted the asymmetry for three decades: we read "I think the market is going up," but never and eight of my last thirty predictions were right. No other field escapes its batting average so completely. The celebrated correct call is usually remembered in isolation — precisely because the person who made it was never right that way again, or because so many similar calls by others were wrong and forgotten. Survivorship does the marketing.

Howard Marks' Own Words

Thought Evolution

Related Concepts

Key Memos

Mentioned In

Source: Chian.io — Howard Marks Knowledge Base