Second-Level Thinking: Why Being Right Doesn’t Pay

First-level thinking: this is a good company, so buy it. This is a strong trend, so go long. This is bad news, so sell.

Howard Marks’s objection is not that these statements are wrong. It is that they are almost certainly correct, and that everybody else has also noticed, and that the price already contains the observation.

Second-level thinking asks a harder question. Not is this good, but is this better than the market currently believes.

The difference between those two questions is the difference between being right and being paid, and the arithmetic is unforgiving.

Your accuracy is not an input

Suppose you are a good analyst. You are right about direction seventy percent of the time. Consensus is right sixty-five percent. And, because you read the same information as everybody else, you agree with consensus eighty percent of the time.

When you agree, the price already reflects the view. There is nothing to buy that has not been bought. Your edge on those trades is zero, regardless of whether you turn out to be right.

So the money is made entirely in the twenty percent of cases where you differ.

Agreement is expensive

Hold your accuracy-when-different at a respectable sixty percent, and vary only how often you find yourself disagreeing with the crowd.

The trader who agrees with everybody ninety-five percent of the time, and is right sixty percent of the time when he does not, has an expectancy of one hundredth of an R. He is, functionally, flat. He is also comfortable, well-informed, and popular, and he will not understand why his account does nothing.

The uncomfortable implication: an edge requires being different, and being different requires being wrong in public forty percent of the time. There is no version of this where you are both correct and unremarkable.

What this looks like on a chart

Marks was writing about securities. It transfers exactly.

News trading. The number is good, and price falls. First-level thinking calls this irrational. Second-level thinking asks what was expected, sees that the number was good and the expectation was better, and notices that “good” was never the variable. Surprise was.

The obvious setup. A textbook break and retest, on a chart everybody is watching, at a level everybody has drawn. The pattern’s historical statistics were measured before everybody drew the level. You are not trading the pattern. You are trading a crowded version of it, and the crowd is your counterparty at the retest.

The prop-firm setup. Tens of thousands of funded traders taught the same three models, watching the same session opens, placing stops in the same place. That stop cluster is not a risk-management artefact. It is inventory, and somebody knows where it is. First-level thinking sees a liquidity sweep. Second-level thinking sees the trade that manufactures it.

Sentiment. “Everybody is bearish” is a first-level observation about opinion. The second-level question is whether they are positioned bearishly, because opinions are free and positions are not, and only positions have to be unwound.

Why this is so hard

Because the first-level view is correct, and correctness is the thing you were trained to seek.

The company is good. The trend is strong. The news is bad. Every one of these is true, and holding a true belief feels like the completion of the work rather than the beginning of it. To go further you must hold two things at once: the belief, and an estimate of how widely the belief is held. The second is far harder to observe than the first, and it is the only one that determines your payoff.

And there is a social cost. Second-level positions are, by construction, positions the well-informed people around you think are wrong. They will remain wrong-looking for as long as the consensus persists, which may be a long time.

Second-level thinking is not contrarianism

This is the error the framework invites, and it is worth killing.

Doing the opposite of the crowd is a first-level view with a minus sign. The crowd is usually right, which is why it is a crowd, and reflexively fading it is a strategy with the crowd’s accuracy inverted.

The second-level trader is not looking for the crowd to be wrong. He is looking for the specific, identifiable cases where the crowd’s view is fully priced and slightly mistaken, and he is content to agree with them the other eighty percent of the time, silently, taking no position at all.

Most of second-level thinking is not trading.

The practical form

Write the consensus before you write your view. One sentence: what does the price currently assume? If you cannot answer, you are not ready to have an opinion about it.

Then write the difference. Not “I think it goes up.” Rather: the market assumes X; I believe Y; the trade pays if Y is nearer to true than X.

Count your disagreements. If you take twenty trades a month and every one of them agrees with the obvious read, your expectancy is a rounding error regardless of how good your analysis is.

Accept the discomfort as the fee. An edge that feels obvious is an edge somebody else has already taken.