The only God of trading is probability

Probability is God on my desk

I have been trying to bring the paid section of The Dark Side of the Boom back into my wheelhouse, because frankly I do not have it in me to keep producing sterile, yesterday-style headline-risk commentary just for the sake of feeding the machine. There are only so many ways to say “markets are watching the Fed,” “geopolitics remains a risk,” or “investors await the next data print” before the whole exercise starts to sound like it was assembled in a windowless committee room with the coffee already cold.

That is not why I started writing this blog.

Yes, the site has taken on a different look with the paywall. Yes, some of the deeper trader interpretations of top-tier bank research will sit behind that wall. And yes, because there is now a brand element involved, not every decision is purely mine in isolation. But this piece is not really about paid versus free. Some daily free articles will still get through. The bigger point is that I am trying to move the work back into the lane where I think I actually have an edge: taking the best nuggets from often tedious institutional research, stripping out the boardroom varnish, and translating them into the language of markets, positioning, probability, and trader consequence.

Because at the end of the day, the only god of trading is probability.

Not conviction. Not narrative. Not a favourite economist. Not the chart pattern everyone discovers after the move has already happened. Not the central-bank whisper. Not even the Bloomberg headline that hits your screen with enough urgency to make your coffee jump. The only thing that matters over time is whether you are consistently better than the crowd at assigning odds to uncertain outcomes, sizing those odds properly, and knowing when the price on the screen is paying you enough to take the other side of the world’s anxiety.

That is trading. Everything else is costume.

As many regular readers know, I grew up in the USD/CAD environment of Bay Street, Toronto, before moving into USD/JPY for a Japanese bank, then stepping into the world of commodities and index trading, before eventually circling back into the wild west of FX. That road later took me to OANDA in 2005 during the first great eFX retail wave, after a stint in Tokyo, where I helped set up trading operations and later ran the book out of Singapore for the next 15 years. It was a strange and formative period: bank-market muscle memory colliding with retail technology, price discovery moving faster, spreads compressing, platforms scaling, and the old dealer world starting to realize that the machines were no longer just tools. They were becoming the arena. Bloomberg and Reuters terminals were no longer just pieces of institutional furniture but the operating system of the market mind. Suddenly, every trader, strategist, sales desk, fund, macro tourist, and fast-money player was drinking from the same hose. Information was no longer scarce in the old way. Judgment became the scarcity.

But in some ways, all of that pales in comparison to what happened when I left that world

That is why an old AM/FX note from Brent Donnelly hit differently. Brent is now at Spectra Markets ( Author of the must-read The Art of Currency Trading and Alpha Trader: The Mindset, Methodology and Mathematics of Professional Trading), but at the time he was writing from inside the big-bank FX ecosystem, where Trader market notes were not trying to win literary prizes; they were trying to help traders survive the next turn in the tape. Brent is Canadian, went to the same university as I did, and although our paths never really crossed, we have exchanged an email or 2 over the years. But buried in that note was a reference to Philip Tetlock, the Canadian forecasting scholar best known for his work on expert judgment and superforecasting, and that single thread pulled me away from the daily market noise and back toward the real operating system of trading.

Probability.

Philip Tetlock’s work ( Superforecasting: The Art and Science of Prediction) matters because it cuts straight through the illusion that markets are mostly about sounding smart. His research showed that expert forecasting is often far messier than the expert class likes to admit, and that the best forecasters are usually not the people with the boldest grand theory of the world. They are the ones who update. They break big questions into smaller ones. They think in probabilities rather than certainties. They are willing to move from 55% to 62% when the evidence changes, rather than waiting for some thunderbolt of confirmation. They track their errors. They learn from misses. They do not confuse confidence with accuracy.

That is exactly how good trading works.

The market is not a morality play. It is not a debate club. It is not a courtroom where the best argument wins. The market is a probability machine with a bad temper. It rewards people who can estimate odds under incomplete information, update without ego, and survive long enough for the edge to matter. It punishes people who confuse a strong narrative with a strong probability.

That distinction is everything.

The tricky thing with market probabilities is that, unlike probabilities in a casino, they are not known. In blackjack, roulette, or dice, the structure of the game is fixed. The distribution may hurt you, but at least you know the machine you are playing against. Markets are different. The wheel changes shape while it is spinning. The dealer gets new information. The rules shift when central banks panic. The crowd changes its mind. Liquidity disappears exactly when your model needs it most. The distribution is not carved into the table. You have to estimate it in real time.

That is why probability estimation is the trader’s real craft.

Anyone can have a view. Views are cheap. The market is drowning in them. The hard part is assigning the correct probability to that view and then comparing it with the price being offered. A trader is not paid for saying “oil can go higher.” A trader is paid for knowing whether the market is pricing a 30% chance of disruption when the real probability is closer to 45%, and whether the option structure, futures curve, liquidity profile, and positioning backdrop make that gap worth expressing.

That is the difference between commentary and edge.

Most market forecasts begin, whether people admit it or not, with the current price and the recent trend. If something is going up, the street usually finds reasons it can go up more. If something is falling, the research machine discovers fresh reasons it can keep falling. This is not because analysts are stupid. Many are extremely smart. It is because humans are wired to overweight recent information. The thing that just happened feels more important because it is available, emotional, visible, and easy to explain.

That is extrapolation bias, and markets are built on it.

When oil is at $100, the world suddenly has 100 reasons it is going to $130. Inventories are tight. Geopolitical risk is rising. Spare capacity is thin. Demand is resilient. The curve is backwardated. The refiners are screaming for barrels. Every argument sounds sophisticated because the price is already confirming it. Then oil trades at $25 and the same world discovers 100 reasons it is going to $15. Storage is full. Demand is dead. Producers are irrational. The curve is broken. The old bullish thesis is not revised; it is buried.

The market does not just change price. It changes the story people are willing to believe.

That is why the best traders are usually not the loudest forecasters. They are the best updaters. They live in probability space, not certainty space. They know when the base case is still alive but the distribution has widened. They know when a 20% tail is being priced like a 5% tail. They know when the market has fallen in love with a clean story and stopped paying attention to the messy alternatives.

This is where Tetlock’s world and the trading floor meet.

The amateur wants the market to tell him what will happen. The professional wants to know what is already priced, what is mispriced, and what happens if the crowd has to change its probability map in a hurry. That is why a 60% view can be a great trade and a 90% view can be a terrible one. If the 60% outcome is priced at 35%, you may have edge. If the 90% outcome is priced at 99%, you may have nothing but crowded certainty and bad convexity.

Price is the bookmaker. Probability is the religion.

This is also why the middle of the probability distribution is easier than the tails. Humans are generally more comfortable estimating things around 40%, 50%, or 60%, because the consequences of being wrong are psychologically softer. The harder task is estimating very high and very low probabilities. Is something really a 5% risk, or is it 15%? Is something really a 95% certainty, or is it 80% dressed up in a good suit? That difference sounds academic until you trade it. In markets, the tails are where portfolios blow up, hedges suddenly matter, and the crowd discovers that “almost impossible” was just another way of saying “underpriced.”

This is where traders often get into trouble.

A low-probability event is not the same thing as an irrelevant event. A 10% risk can dominate the entire trading landscape if the payoff is large enough, the market is positioned the wrong way, or the event path forces liquidation. Likewise, a high-probability event can be useless if everyone already owns it, the price has already moved, and the residual upside is tiny compared with the downside if the consensus is even slightly wrong.

That is why I spend so much time thinking about tail risks.

Not because every tail risk happens. Most do not. That is why they are tails. But markets do not wait for the tail to happen before they move. They reprice the probability of the tail. That repricing can be enough. Oil does not need the Strait of Hormuz to shut completely for crude to carry a risk premium. USD/JPY does not need the Bank of Japan to shock the world for traders to start paying up for a wider distribution around the meeting. Gold does not need the monetary system to collapse to rally when real yields fall, fiscal credibility erodes, and investors start marking a higher probability of policy error.

Markets trade probability before they trade fact.

That is also where crowd-sourced probability tools, prediction markets, and places like Polymarket become interesting. I am not saying they are gospel. They are not. They have their own liquidity issues, participant biases, domain distortions, and moments where the crowd is simply leaning too far into its own reflection. But they are useful because they force the conversation into numbers. Not vibes. Not “likely.” Not “could.” Not “watching closely.” A price-based probability forces you to ask: what odds is the crowd actually assigning to this outcome?

That alone is valuable.

Markets are full of language designed to avoid accountability. Strategists say risks are “elevated.” Economists say the balance of risks is “skewed.” Central bankers say policy is “data dependent.” Traders say something “feels heavy.” None of that is useless, but none of it is a probability. The discipline comes from forcing the vague into the numerical. Is this a 20% risk or a 40% risk? Is the market pricing 2 cuts when the real distribution says 1.3? Is the crowd paying for insurance after the house has already stopped smoking?

That is where edge starts to appear.

But probability is not enough by itself. You also need calibration. A trader who calls everything 70% is not a forecaster; he is a man with one dial. A strategist who constantly predicts dramatic outcomes may occasionally look brilliant, but without scoring the misses, there is no way to know whether the process has any value. The same is true of the market itself. A contract priced at 70 does not automatically mean the real-world probability is exactly 70%. It means the marginal price, under that market’s liquidity, participation, constraints, and incentives, is 70. That distinction matters.

The crowd can be wise, but it can also be crowded.

This is where Tetlock’s world and trading meet perfectly. Good forecasters track their errors. Good traders do the same, although usually in the harsher language of P&L. If your 60% trades win 60% of the time over a meaningful sample, you are calibrated. If your 80% trades win 55% of the time, you are not unlucky; you are miscalibrated. The market is giving you feedback. The question is whether you are humble enough to listen before the drawdown becomes your teacher.

That is why probability is not cold or academic to me. It is survival.

Every trader eventually learns that the market does not care how much work you put into the view. It does not care how many bank notes you read, how many charts you marked up, or how elegant the thesis sounded when you explained it at 6 AM. The only thing the market cares about is whether the odds, price, timing, and size were aligned. You can be directionally right and lose money. You can be intellectually right and structurally wrong. You can spot the macro turn and still get carried out because the entry, carry, volatility, or positioning was wrong.

Probability is not just about being right. It is about being paid correctly for being right and surviving when you are wrong.

That is why I keep coming back to the same idea: markets are not prediction contests. They are probability contests with leverage attached. The objective is not to have the most dramatic forecast. The objective is to find where the market’s implied odds are wrong, where the crowd is extrapolating too aggressively, where the tail is underpriced, and where a small change in probability can force a large change in price.

That is the real game.

And maybe that is where this blog is heading again. Less sterile headline risk. Less yesterday’s consensus dressed up as today’s insight. More probability. More trader implication. More of the messy middle where institutional research, price action, crowd behaviour, and tail-risk thinking collide. Because that is where I have always felt most at home.