What Survivorship Bias Looks Like From Inside

I have held through several severe declines and I am still here writing about it. That fact is doing more work in the credibility of everything I write than any of my reasoning.

The selection

The people writing about long-term holding are the ones whose long-term holding worked.

The ones who held through 2018, exhausted themselves, sold near the low and never came back do not write about it. The ones who held something that never recovered stopped having a topic.

So the visible sample of long-term holders is filtered by outcome, and I am in the filtered sample.

What that means for my advice

When I write that patience works, I am generalising from a sample that consists of people for whom it did.

The honest version is narrower: patience worked for me, holding these assets, across these specific cycles. Whether it works in general is not something my experience can establish, because my experience is conditional on it having worked.

The version of this that is genuinely misleading

Someone who bought early, held through everything, and now has a substantial position writes about conviction.

The conviction is real. It is also inseparable from the outcome, and there is no way for that person, or for me, to know how much of it was judgement and how much was being right about which asset.

The same conviction, applied to an asset that did not recover, produces a story nobody publishes.

What I do about it

Write about method rather than outcome. A rebalancing rule can be evaluated on its logic. A return cannot be separated from luck.

Publish the failures. The position I exited, the quarters where I answered questions from memory, the arguments I found compelling and did not act on. Those are the parts of the record that are not selected by success.

Score predictions. Nine right, eight wrong, six too vague. That number is not flattering and it is not filtered.

State the conditionality. Everything I conclude is conditional on having held assets that recovered. Someone applying the same method to something that did not would have the same method and a worse result.

The part that survives the objection

Some things are true independent of outcome.

A restore test either catches a transcription error or it does not, regardless of what the price did. An inheritance document either works or it does not. A position sized so that a seventy percent decline cannot force a sale is safer than one that is not, whether or not the decline happens.

Those are the parts I am most confident about, and it is not a coincidence that they are the operational parts rather than the judgement parts.

The rest, including everything I have written about conviction and patience, should be read as the account of someone who is here because it worked, which is exactly the population you should be most careful about learning from.