Alternative histories

Something happened, and then you judged it. The trade worked, so it was a good trade. The bet paid, so it was a smart bet. This is the most natural reading in the world, and for most of daily life it is roughly fine. In markets it is quietly poisonous, because in markets the thing you got to see — the outcome — is only one of the many things that could have happened, and often not even the likely one.

Picture a friend who put most of his savings into a single leveraged position and walked away with a large gain in a week. The whole family now speaks of him as someone who understands the market. But you and he saw exactly one version of events unfold. There were other versions — histories that were fully available to the same bet, in which the position moved the other way and the savings were gone. Those versions did not happen. That does not mean they were not there.

This module is about learning to see them anyway — the alternative histories, the paths the world could have taken but didn't. The idea is Nassim Taleb's, from Fooled by Randomness, and it is one of the most useful mental habits a beginner can build, because it attacks the single error underneath most reckless investing: judging a decision by the outcome that landed instead of by the full spread of outcomes it exposed you to.

Here is the shift in view, and it is a large one. When you make an investing decision, you are not choosing an outcome. You are choosing a bet — a spread of possible outcomes, each with some chance of occurring. Then the world runs the bet exactly once and hands you a single result. The result is real. But it is a terrible judge of the bet, because it is a sample of one drawn from a distribution you never get to see in full.

A good decision is one that comes out well across most of that distribution — that holds up in the histories that didn't happen as well as the one that did. A bad decision is one that needs the world to take a lucky path to survive. And crucially, you cannot tell these two apart by looking at the outcome alone. A good decision can land badly (you did everything right and got unlucky). A bad decision can land beautifully (you were reckless and got rescued by chance). The person who bet the savings and won made a bad decision that happened to pay — the outcome flattered a bet that most of its own histories would have punished.

This is why outcome bias — grading the decision by the result — is so dangerous here specifically. It rewards exactly the behaviour that will eventually ruin the person doing it, because it keeps handing praise and confidence to bets that got lucky, right up until the history that doesn't spare them finally arrives. The market is a slow machine for teaching people the wrong lesson, one flattering outcome at a time.

The repair is a change of question. Not "did it work?" but "across the paths that were available, how many would this decision have survived?" You will never know the true distribution. But even a rough, honest sketch of it — a few plausible paths written down — is enough to tell a survivable bet from a fragile one, and that is most of the battle.

The reason alternative histories matter so much more in markets than in, say, a coin game for matchsticks is a single hard asymmetry: some paths are permanent. This is path-dependence — where you can end up depends on the road you actually travelled, not just on the odds you faced along the way. You can recover from most bad outcomes. You cannot recover from ruin — the outcome that takes you out of the game entirely, wipes the capital, ends the account. And a path that leads to ruin does not care that it was unlikely, because once it happens, none of the good paths that came after are available to you anymore.

This is the flaw in the most seductive-looking track records. Consider a strategy that wins nine times and blows up on the tenth. Each win looks like evidence; the streak looks like skill; the confidence compounds. But if the tenth outcome gives back everything and more, the strategy is a losing one — it was losing all along, and the nine wins were simply the part of the path before the loss arrived. The record was never hiding skill. It was hiding a tail.

There is a slightly technical name for the reason your intuition fails here, worth meeting once. Your gut assumes that if a bet is good "on average," repeating it is safe. That is only true when the average across many people at once equals the average across one person over time — a property called ergodicity. Bets with a ruin path are not ergodic: the average across a thousand players can look wonderful while the typical single player, living the bet one round after another, goes broke — because that player, unlike the crowd, only has to hit the ruin path once to be removed for good. The average includes lives that are no longer possible for you.

The practical name for the danger is tail-risk: a rare, large, sometimes irreversible outcome sitting in the far edge of the distribution — small in probability, enormous in consequence. Most of the histories are unremarkable. The one that ends you is out in the tail, and it is precisely the one that a run of good outcomes trains you to ignore.

It is one thing to read this and nod, another to watch it happen. The toy below runs the same bet across fifteen hundred lives. Every round, most of the time it pays a modest gain; rarely, it takes a large loss. On paper the average round can look positive — genuinely tempting. illustrative

Watch the two numbers that matter pull apart. The average life can end up handsomely, dragged up by a few explosive winners — that is the figure a brochure would quote. But the typical life, the one in the middle that you are most likely to actually live, can be shrinking at the same time; and a real slice of lives fall through the ruin floor and stop, never getting the good rounds that came later. The winner you'd read about online is one of the top green lines. The red graveyard never posts.

Now bring it back to a single ordinary decision. A reader is sizing a promising small-cap and reaches, naturally, for the outcome he expects — it doubles in two years — and sizes the position around that. The mistake is invisible to him, because the good path is the only one he is looking at.

The repair is not cleverness or a forecast. It is base rates before stories applied to your own bet: before acting, ask what usually happens to positions like this one, sketch two or three honest paths, and size against the bad one — not the one you are hoping for.

The whole discipline collapses into one small, calm habit you do before you act, while you can still think clearly. Write down two or three plausible histories for the decision in front of you — not a forecast, just an honest spread:

1. The good path. The thesis works; things go roughly as you hope.
2. The dull path. Nothing much happens; the idea neither breaks nor rewards.
3. The bad path. Something specific goes wrong — a fraud comes to light, a fund-raise dilutes you, the sector de-rates, the price halves.

Then size the position so that the bad path is survivable — so that if the history you didn't want is the one the world takes, you are bruised but still in the game, still holding capital, still able to walk the good paths that come later. This single move quietly defuses almost every reckless decision, because recklessness is nearly always the act of sizing to the good path and pretending the bad one isn't there.

This habit is powerful, and like every powerful habit it has edges where it stops working, and pretending otherwise is its own trap.

It does not let you count the paths. You will never know the true distribution of outcomes, only a rough, human sketch of it. The sketch is enough to separate a survivable bet from a fragile one — but it cannot be dressed up as a precise probability, and a made-up percentage is not more honest for having a decimal point.

It is not an argument for never taking risk. Sizing everything for the worst imaginable path leads to a different failure — owning nothing, compounding nothing, being ruined slowly by caution instead of quickly by recklessness. The goal is not to eliminate the bad path; it is to make it survivable, so that you are still standing to take the next good bet.

And it can be twisted into hindsight cleverness. After a loss, it is easy to say "obviously that path was there" — but if you never wrote the paths down before, you are not thinking in alternative histories, you are just editing your memory to look wise. The discipline only counts when the paths are written before the world runs the bet, not narrated after.

The same handful of moves catch beginner after beginner. Named once, they are far easier to catch in yourself.

1. Reading a good outcome as a good decision. The result is one sample from a distribution you never saw. A win can flatter a reckless bet; a loss can punish a sound one. Grade the bet, not the sample.

2. Treating a winning streak as proof of safety. A run of good outcomes is exactly what a hidden tail looks like from the inside, right up until it isn't. The absence of the bad path is not evidence that there is no bad path.

3. Dismissing a risk because it didn't happen. "Nothing went wrong" describes the outcome, not the odds. A risk that was real beforehand stays real even when the world spares you.

4. Sizing to the hoped path. The single most common route to ruin is a position sized for the outcome you want, with the bad outcome left unpriced. Size for the history you'd hate, not the one you'd love.

5. Admiring the survivor without counting the graveyard. The person who bet everything and won is visible; the many who made the same bet and were wiped out are not. Copying the survivor copies the exposure, not the luck.