Thinking in Probabilities

intermediate6 min read

Any single trade is noise; the edge shows up over hundreds. Why you must zoom out.

A positive-expectancy edge is a statistical truth that only reveals itself over many trades. On any single trade, anything can happen — a great setup loses, a terrible one wins. Learning to think in probabilities, not certainties, is what lets you follow a good system through the noise.

The key mental shift: judge your decisions, not your individual outcomes — because in the short run, luck swamps skill. Like a casino, your edge is real but only materialises over a large sample; a single trade (or even a string of them) is pure noise that says little about whether your system is good. This is the law of large numbers: flip a fair-ish coin ten times and you might get seven tails; flip it ten thousand times and the true odds emerge. So a good trade that loses was still a good trade, and a bad trade that wins was still a bad trade. Traders who judge each result individually abandon winning systems during normal losing streaks and chase lucky junk — sabotaged by outcome bias. Zoom out: trust the process across hundreds of trades, not the verdict of any one.

Single trades are noise — even a strong edge loses often; one result tells you almost nothing.
Law of large numbers — the edge (and your true expectancy) only emerges over a large sample of trades.
Judge decisions, not outcomes — a good process can lose a trade; outcome bias makes people quit winners and chase luck.

ExampleA casino can lose to a lucky gambler all night, yet knows it will profit over millions of bets — it never panics over one table. A systematic trader with positive expectancy thinks the same way: a 6-trade losing streak is expected variance, not proof the system is broken. Quitting there is judging noise as signal.

Key takeawayA real edge only shows up over many trades (the law of large numbers); any single trade is noise. Judge your decisions/process, not individual outcomes — a good trade can lose and a bad one can win. Zoom out and trust positive expectancy across hundreds of trades, not the result of one.

FAQs

How many trades before I trust (or reject) a strategy?

Generally far more than people expect — dozens prove little; you often need hundreds for results to be statistically meaningful (covered in the metrics module). Small samples are dominated by luck, which is why both abandoning a system after a few losses and trusting one after a few wins are classic mistakes. Demand a large sample before concluding.