Persistence and learning
Marcus recounts early losses, learning from others and gradually developing independent judgement. The chapter does not support the simple claim that persistence alone creates success. Repeated failure becomes informative only if exposure, reasoning or execution changes. His account of holding useful positions also balances the familiar emphasis on cutting losses: a process must consider both sides of its outcome distribution.
Our interpretation is to translate borrowed advice into a claim that you can explain and test. If you cannot say what invalidates a tip, you do not yet have a complete decision. Nor should one apparent agreement between price and fundamentals erase uncertainty. The process needs a boundary around how much can be lost while the idea is evaluated.
Worked example
Four fictional outcomes are −1, −1, −1 and +5 units: total +2 before costs. Cutting the last outcome to +1 changes the total to −2. This arithmetic illustrates payoff shape, not a forecast of a strategy.
Limits
The survivors in an interview collection are a selected group. Repeatedly replenishing a losing account is not evidence of learning, and dedication cannot guarantee a profitable method.
Case connection
The currency case tests whether an apparently reliable opinion rests on a policy assumption. Independent judgement starts with making that assumption explicit.
The Swiss franc: when the boundary moves
A policy boundary is a decision made by an institution. It is not a permanent physical property of the market.
The documented policy change
On 15 January 2015, the Swiss National Bank discontinued its minimum exchange rate of CHF 1.20 per euro and lowered the interest rate on sight-deposit balances to −0.75%. In April, Thomas Jordan explained that maintaining the boundary had required interventions of rapidly increasing size as the euro weakened. The SNB described the old arrangement as unsustainable. [1, 2]
What changed, and what did not
The April explanation also said the SNB would remain active in foreign-exchange markets when necessary. Ending a particular boundary did not mean abandoning monetary policy. This matters because a binary story—either a fixed promise or complete inaction—misses the institution’s ability to change instruments. The public explanation gives a rationale for the decision, but does not establish what every market participant knew beforehand. [1]
Interpretation: a model can hide a policy bet
Consider a hypothetical process calibrated to a period when an exchange rate rarely crosses a threshold. The process may appear to identify stable statistical behaviour. Yet part of that stability can depend on an institution continuing to intervene. If the model treats the observed boundary as a law rather than a policy choice, its estimate of risk quietly embeds a political and operational assumption. More historical observations under the same arrangement do not necessarily diversify that assumption.
The unit of risk is the shared dependency
Kovner’s portfolio perspective is useful here. Several fictional positions can have different labels and still depend on the same currency arrangement. Counting tickets is not the same as counting independent risks. A proper classroom stress exercise asks what happens to all positions if the common assumption fails. It should also distinguish cash already available from assets that must first be sold or transferred. A funding plan is only useful if resources arrive when obligations fall due.
A hypothetical gap test
Suppose a classroom position has 10,000 units of exposure and the planned loss is one unit per unit of exposure. The modelled loss is 10,000. If the next executable exit is three units away, the corresponding loss is 30,000 before costs. These numbers are invented, not observations of the Swiss-franc event. Their purpose is to show why a planned stop and a maximum possible loss are different quantities. Increasing leverage does not make the policy assumption more reliable; it magnifies the consequences of being wrong.
Avoid learning the opposite superstition
The wrong response is to conclude that every official statement is worthless or that every policy boundary must collapse immediately. Both claims replace analysis with a slogan. The useful response is to identify the institution’s objective, the tools available, the costs of continuing the policy and a scenario in which the policy changes. Even that analysis cannot provide a reliable date. It can make an otherwise invisible dependency explicit and help frame an appropriately limited educational model.
The memorable lesson
Ask whether apparent stability comes from independent market forces or from an arrangement that somebody can revise. Then ask whether the strategy remains coherent outside that arrangement. This case supports a critical reading of confidence in the interviews: confidence is not certainty about a forecast. It is clarity about what would count as a changed environment, what can be controlled and what remains outside the participant’s control.
Consider
Which apparently stable feature of your model depends on someone else continuing to act?
Analysis guide
Name the institution or mechanism, the assumed action, and a plausible change. Test the whole fictional portfolio under that change. Avoid assigning a probability or exact historical fill without evidence.
SNB · Thomas Jordan, 24 April 2015 · SNB · Announcement, 15 January 2015
Reflection
What have you changed after a mistake, and how will you check it helped?