Risk & Psychology · intermediate · 8 min

Common Biases

The hardest thing to be honest about in trading is yourself. The mind runs a set of shortcuts that were useful on the savannah and are actively expensive in markets — they make randomness look like a pattern, make a losing approach feel like it's working, and rewrite your reasoning after the fact so you never quite learn from it. They are not character flaws; they are standard-issue and universal. Knowing their shape is the first, cheapest defence. This lesson walks through the ones that cost traders the most.

Recency bias

Recency bias is the tendency to weight the last few observations far more heavily than they deserve. Three winning trades in a row and the strategy feels unstoppable; three losers and it feels broken. In both cases the sample is three, which tells you almost nothing — but it feels like everything because it just happened.

The damage is concrete: recency bias pushes you to size up right after a hot streak (exactly when regression to the mean is most likely) and to abandon a sound approach during a normal losing run (exactly when quitting locks in the loss). The antidote is sample size. A win rate over 200 logged trades means something; a feeling about the last five does not. This is one of the reasons a journal matters — it replaces the vivid recent memory with the full count.

Confirmation bias

Confirmation bias is seeking, noticing, and believing evidence that supports what you already think, while skimming past evidence that contradicts it. Having decided a market is going up, you read every green candle as proof and explain away every red one. On a chart you can always find something supporting the view you arrived with, because charts are rich enough to support almost any story after the fact.

This is the bias that makes backtesting so treacherous. If you go looking for a rule that "worked" on the history, you will find one — the past is large and noise is patient. That found rule is usually confirmation bias wearing the costume of research. The defence is built into how Visor tests strategies: the random control asks whether a rule beats a scrambled version of itself, which is a question confirmation bias cannot talk its way around. See The Overfitting Trap for the full treatment.

Hindsight bias and the stories we tell

Once an outcome is known, the mind instantly rewrites the past so the outcome looks as though it was obvious all along. A trade taken on a hunch, if it wins, is remembered as a trade you "read perfectly." The chart, in hindsight, looks like it had to go that way — the entry looks clean, the exit looks timed, the uncertainty you actually felt is deleted.

This is hindsight bias, and it is corrosive because it destroys the feedback loop. If every winner was "obviously" skill and every loser was "just bad luck," you can never tell a good decision from a good outcome — and in a probabilistic game those are different things. A well-reasoned trade can lose and a reckless one can win; judging decisions by their outcomes rewards the reckless ones whenever they happen to pay off. The only real defence is a record written before the outcome is known, which is precisely what a trade journal is and why its entries are timestamped at entry, not filled in afterwards.

A few more worth naming

Loss aversion — a loss hurts roughly twice as much as an equivalent gain feels good. This is why traders cut winners early (to lock in the good feeling) and let losers run (to avoid crystallising the pain) — the exact opposite of what the drawdown maths in Drawdown and Recovery rewards.
The gambler's fallacy — believing a run of losses makes a win "due." Independent trades have no memory; the market does not owe you a green one for sitting through five reds.
Overconfidence after a win — a good result breeds the certainty that leads to oversizing the next trade, which is where Risk of Ruin starts.

The honest takeaway

You do not out-think these biases by being smart — the smart are, if anything, better at constructing convincing rationalisations. You beat them with process: a record written before outcomes are known, a sample size large enough to mean something, and tests that a story cannot argue with. That is the whole reason this track pairs psychology with arithmetic and with the robustness gates. The biases are the reason the honest tools exist; the tools are how you trade despite the biases rather than pretending you're above them.