Confirmation Bias in Trading: How It Distorts Analysis

Confirmation bias in trading: how it turns analysis into justification, the three places it operates — before entry, during a trade, and in review — and the procedural checks that catch it.

Confirmation bias is the tendency to notice, weight and remember evidence that supports what you already believe, and to discount evidence that does not. In trading it does not feel like bias. It feels like analysis: the chart agrees, the indicator confirms, the news makes sense. The problem is that a trader who wants to be long can find all three for any market on any day.

Key Takeaways

• Confirmation bias operates in three places: choosing the trade, holding the trade, and reviewing the trade.
• With a dozen indicators available, one always agrees. That is not confirmation; it is selection.
• The bias edits memory: the setups that worked are remembered, the identical ones that failed are not.
• The fix is to look for the disconfirming case deliberately and to keep a record that cannot be edited.

Before Entry: Analysis as Justification

The sequence usually runs: a view forms — from a headline, a chart glance, a previous trade — and then evidence is gathered. The gathering is where the bias lives. A trader who wants to buy EURUSD checks a moving average (bullish), a momentum indicator (bullish), and the calendar (nothing bad scheduled). The two indicators showing the opposite were not consulted, or were consulted and discounted as "lagging".

The test: what would have to be true for this trade to be wrong, and have I looked? If the answer is "nothing I can think of", the analysis has not been done; the conclusion has been decorated.

During the Trade: Reading the Chart for Reasons to Hold

A position at −$30 against a −$30 stop generates a search for reasons the level will hold. A support line that was not on the chart yesterday appears. A news item is read as supportive. This is loss aversion recruiting confirmation bias to justify moving the stop — and it is why the stop belongs on the server, where the search for reasons cannot reach it.

In Review: Editing the Record

The most expensive form. Over a month, the setup that worked three times is remembered as a good setup. The same setup that failed four times is remembered as "different conditions". Without a written record, the review confirms the strategy that the review was meant to test.

Worked Example: Indicator Shopping

EURUSD at 1.0850. Trader wants to buy.

One of five supports the trade; three of five argue against it; one was reframed. The trade is taken on the one. This is not analysis with a bullish result. It is a bullish result with analysis attached afterwards.

(Illustrative.)

Procedural Fixes

• Write the disconfirming case first. Before the entry: one line on what would make this trade wrong, and whether it is present. See technical vs fundamental analysis for the distinction between evidence and support.
• Fix the indicator set in advance. Three indicators, chosen when no trade is in view, consulted every time. Adding a fourth mid-analysis is shopping.
• Define the setup in writing — the exact conditions — and score each trade against the definition, not against the outcome.
• Journal the setup name and the plan compliance, not the result. See how to keep a trading journal. Review by setup, over a sample, and let the count decide.
• Backtest with the rules fixed before the data is seen. Adjusting rules to fit the past is confirmation bias with a spreadsheet — see how to backtest a trading strategy.

Why It Matters

Confirmation bias is the reason a losing strategy can feel like a winning one for months. Every trade had reasons; every loss had an explanation; the record was never kept in a form that could disagree. It is also the reason market sentiment analysis is so easily abused: with a dozen sentiment measures, one always supports the position.

How is confirmation bias different from having a view?

A view is a starting point. Confirmation bias is what happens when the evidence-gathering is shaped to protect it. The test is whether disconfirming evidence was sought.

Do indicators cause confirmation bias?

The number of them does. A fixed set consulted every time is analysis; a growing set consulted until one agrees is selection.

Can backtesting remove it?

Only if the rules are set before the data is examined. Fitting rules to the past is the same bias in a different form.

Is it worse in fundamental or technical analysis?

Both. Fundamental narratives are easier to rationalise; technical indicators are easier to shop. Neither is immune.

What is the single most useful check?

One written line, before entry: what would make this wrong, and is it here?

Related Reading

Trading psychology: how emotions affect decisions · Loss aversion · How to read market sentiment · Technical vs fundamental analysis · How to keep a trading journal · How to backtest a trading strategy