3. When to cut for real, and how not to confuse it with fear

Risk 9: A risk system of your own

This is the hardest module in the course and the one that decides the most money. The question looks simple and is not: how do you tell a justified reduction from a flight dressed up as prudence?

The two legitimate reasons to cut

There are only two, and neither mentions the recent price.

The first is that the thesis has died. Not that it is going badly: that the thing you yourself wrote down as what would make you change your mind has happened. If you never wrote it down, you have no way of knowing whether it has happened, and any fall will be reinterpreted as an opportunity and any rise as confirmation of talent.

The second is that the real risk has risen, whether you decided it or not. This covers very different things: that the weights have moved on their own and one position now takes up twice the room; that effective leverage has grown because the asset fell and the debt did not; that the instrument has changed structure; that the vehicle's liquidity has deteriorated; that a counterparty has appeared where there was none before; or that you have changed, because you lost your job, your time horizon shortened or you started withdrawing money.

Notice that the second reason is checked with data that sits in your holdings statement and in the product documents, not on the chart. And notice what appears in neither of the two: that the position is showing a loss. Being down is not information about the future of that position, it is information about your past, and level 2 already explained that the purchase price plays no part in any sensible position sizing decision.

Criteria that do not depend on the recent price

• Has any of my written limits been crossed, measured by exposure and not by price?
• Has anything happened from my list of what would make me change my mind, drafted before buying?
• Has the structure of the instrument changed: embedded leverage, counterparty, custody, liquidity of the underlying, tax regime?
• Has my capacity to hold on changed: income, payment commitments, time horizon, withdrawal rate?
• With new money, and without recalling my purchase price, would I put on this position at this size today?

The last one deserves a qualification. It is the only one that touches on price, and it works because it neutralizes the anchor to the acquisition cost, not because the price is information. If the answer is no and you cannot explain what has changed, the problem is not in the position, it is that you bought it without any criterion in the first place.

How to recognize fear when it is in disguise

There are fairly reliable signs that you are cutting out of pain rather than out of analysis:

• Urgency. Legitimate reasons hardly ever demand execution today; fear always does.
• The size of the decision is out of proportion to what you claim to have discovered: you go from 100% to 0%, not from 8% to 4%.
• The reduction comes after the move and not after the data, and coincides with the day the headline was biggest.
• You could not write the decision out in three lines without mentioning how much you have lost.

Annie Duke insists on a point that is central here: judging a decision by its outcome, what she calls resulting, makes learning impossible. Selling in a panic and being right is still a bad decision, because the procedure that produced it cannot be repeated.

The symmetrical error: holding on out of pride and calling it discipline

Popular financial culture has turned "holding on" into a virtue, and that generates the opposite error, quieter and often more expensive. Signs that you are not being disciplined but stubborn:

• You cannot state, today and without rereading anything, what specific and observable event would make you close the position.
• You have been buying more on every fall with no new information, purely on price.
• The arguments you now defend it with are not the ones you had the day you bought it: the thesis has been rewritten as it failed.
• You started calling it a long-term investment right after it started going badly.
• You have stopped measuring it, because measuring it means having to look.

Philip Tetlock described the two failures as sides of the same coin: some never update, anchored to their first opinion, and others update on any headline. Good forecasters update often, in small increments and only on evidence. Howard Marks adds the distinction that is hardest to accept: being wrong and being early feel very much alike from the inside, and only criteria written down beforehand let you separate them honestly.

A real case: February 5, 2018

For years there existed an exchange-traded note that bet inversely on volatility: it made money as long as the market stayed calm. It worked exceptionally well over a long stretch and came to hold more than a billion dollars. Its prospectus contained a very specific clause: if the indicative value fell by around 80% in a single session, the issuer could wind the product up early.

On February 5, 2018 the US volatility index rose in one day from levels near 17 to around 37, the biggest daily percentage jump in its history. The value of the note fell around 96% after the close, the clause was triggered and the product was liquidated weeks later. Whoever held it did not lose part of it: they lost practically all of it, with no chance of waiting for the market to normalize.

What matters for this module is when it was right to cut. Not on February 5, when the price was already on the floor and the decision was impossible, but months earlier, reading the prospectus: there was a clause that turned a big fall into a permanent loss. That is a structural change in risk, checkable without looking at the chart. Whoever sold in January because the product's rise made them nervous got the outcome right; whoever sold because they had understood the clause took a decision they can repeat for the rest of their life.