Risk

You are a trader. Not because you touch markets, but because every choice spends one option to buy another: the job you took cost the three you turned down, the evening you spent here cost everywhere else you could have been. You do it constantly, under varying levels of uncertainty and risk, betting that what you do is worth more than what you gave up.

Risk is exposure to the possibility of loss, accepted in pursuit of gain. You take the potential for downside because you want the upside. Two things must be present:

1. Uncertainty: the outcome is not settled, otherwise you have a known cost.
2. Stakes: something you value is on the line, otherwise the uncertainty is just trivia.

Two people can take the same position and perceive different risks, because uncertainty lives partly in the gap between what is happening and what you understand. Obtain a superior understanding and your bet gets safer, not because the world changed but because your model of it did.

There are two errors that are made when it comes to taking risks:

1. Excess risk: you stake more than you should, so one bad outcome could cripple you. A more disciplined plan arbitrages your recklessness.
2. Insufficient risk: you stake too little, and the upside goes unclaimed. A bolder person arbitrages your timidity.

The whole craft is sitting between them.

Where predictability comes from

A single person is impossible to model because of free will; a crowd of them is much easier to model. One spin of a roulette wheel is anyone's guess, but the casino banks on a million spins and knows its nightly take to the dollar, because each spin is independent of the last and the house edge is iron across the aggregate. Unpredictable parts, more predictable whole. Casinos, insurers and index funds live here.

It rests on one assumption: the parts stay independent. When decisions stop moving separately and start moving together, the law of large numbers collapses and the tails fatten. This was 2008: mortgages were priced as far more independent than they were, then one shock made them fail at once, and a "diversified" pool revealed itself to be a single correlated bet.

What makes independent parts synchronise is people reacting to each other. Reflexivity: I act on what I think you will do, you on what you think I will do, until our choices lock together. Reflexivity is a correlation-generating engine, and it is why markets and panics are so much harder to model than dice. So never assume independence; in any system with people in it, hidden correlation is the default.

None of this is equal. The same system, thick with hidden correlation, is chaos to one person and an open book to another who understands it better. Predictability tracks understanding, so where your grasp outruns the other side's, that gap becomes your edge. This is information asymmetry, and there is nothing underhanded in it: it is earned through privileged information, better analysis, or a sharper read of reality. When Soros broke the Bank of England he had the same information as everyone else, only a truer read: the peg was unsustainable and the Bank politically bound to defend it anyway. Reality paid him for understanding it better than the people on the other side. Edge is better information and a better model, and most often it is the model that decides, because information is far easier to gather than understanding.

Reality keeps score

You can never measure risk exactly; if you could, it would simply be a cost, not a risk. So you do not trade in certainties, you trade in distributions, your best estimate of the odds. You cannot be right about a single probabilistic outcome, only well-calibrated across many.

That is what makes reality the only judge. Over enough attempts the rewards flow to whoever held the most accurate model, even when no model is perfect. You do not have to be right, only less wrong than the other side, again and again.

Know which game you are in

Before sizing any bet, read the board. Your environment sets the boundaries inside which a bet is sane: the same wager is prudent in one regime and suicidal in another. Three questions read it:

1. Is there an anchor to reality, a cash flow or a use? Anchored things drift back toward value; unanchored ones are pure reflexivity and can go anywhere.
2. Are correlations rising toward one? When everything starts moving together, independence is dying.
3. Is the feedback damping or amplifying? Self-correcting systems are mild; self-feeding ones, runs and bubbles, build until they break.

Reading is not certainty. The turkey is fed every day, more confident each time, right up to the afternoon before Thanksgiving; the data was never going to warn it. Calm itself feeds the blow-up, because stability lulls everyone into hidden leverage and crowding. Detection warns, but it never certifies, nothing can replace understanding the root causes.

The deepest error is the wrong map: bringing mild-world (thin tails) tools, averages and bell curves, to a wild world (fat tails) where a single event dwarfs everything else. That was 2008 again, the same crash from the other side: not just blind to correlation, but using the wrong math for the world it was in. And regimes nest, so you only need the one whose tail can reach you; the memecoin trader and the central banker live in different worlds. The non-edge default is the index fund, which is just diversification: protection against ignorance.

How to take risks

Now you can act, and there are two ways to win. The first is to be more right than the other side. When the crowd is overcautious, risk is underpriced and owning it pays; when the crowd is greedy, risk is dear and you sell it to them or stand aside. Their timidity and their greed are both your arbitrage. The second needs no special insight: bear what others cannot. The price can be fair and the danger real, yet you are still paid a premium for holding what a short horizon, thin pockets, or a weak stomach forces everyone else to shed. One edge is being right; the other is being able to endure.

An edge is not enough; you have to size it properly. Bet too little and you waste it; bet too much and variance ruins you before the edge can pay off. The Kelly criterion names the fraction of your bankroll that maximises long-run growth: the most underpriced risk you can take without crossing into the reckless error.

Why not just chase the highest expected value? Because you live one life, not a thousand parallel ones. A coin that pays +50% on heads and loses 40% on tails "earns" 5% averaged across many players, yet a single player betting it again and again goes broke, because losses compound and ruin is permanent. That gap, between the average across worlds and the path through your one life, is what physicists call non-ergodicity. Survival comes first; you cannot compound from zero.

Then let PnL and reality keep score. A single loss is a cast, not a verdict; no fisher reels in every throw. But consistent loss is the world telling you your model is wrong. "I was right but lost money" is almost always a misread: Burry and Taleb bled for years, yet they had priced how long they could afford to and sized to survive it. The waiting was part of the thesis. Being right too early to survive is just a flattering way of being wrong.

Remember: this is not only about money. A loss is just a message from reality: read it, update the model, place the next bet with the benefit of more information and time. The discipline is the same whether the stake is capital, a career, or a person you might love. Almost everyone plays too safe, diversifying their whole life into the index, the steady job, the default option, and calling it prudence. But the index is what you buy when you have no edge, and in your own life you hold an edge no one else can. Insufficient risk has its cost too, paid quietly, in upside never claimed and a life never lived. So take more of the risks you understand. We are all traders.