Cognitive Biases in Trading: 6 Common Biases and How Rules Counter Them
Cognitive biases in trading are predictable mental shortcuts that distort decisions, such as confirmation bias, loss aversion, recency bias, overconfidence, anchoring and sunk cost. Each leaves a pattern you can find in a trade record, and a written rule can counter it.
Module 16 · Lesson 1 of 4
What you'll learn
- What cognitive biases in trading are, and why they are patterns in a record rather than moods.
- Six biases with the research behind each.
- A trading example for each bias, and a written rule that counters it.
- How skilled traders turn bias research into a journal review.
This lesson is part of the Trading Mindset module.
What are cognitive biases in trading?
Cognitive biases in trading are predictable mental shortcuts that push decisions away from your own plan. Researchers described most of them in experiments and surveys, and some have since been measured in brokerage records. They are not character flaws, and a trade journal makes them visible.
The useful way to treat a bias is as a pattern that shows up in a journal. A feeling is hard to measure, but "I moved my stop on four of my last six losers" is a fact you can count, and the trading journal lesson shows how to record it.
Why beginners lose lists the common mistakes, FOMO in trading covers fear and greed at entry, and revenge trading covers the urge to win a loss back. This lesson asks which documented thinking errors sit underneath those behaviours, and what written rule answers each one.
Definitions
- Bias: a systematic, repeatable tilt in judgement, not a one-off mistake.
- Heuristic: a mental shortcut that is usually good enough and sometimes badly wrong.
- Counter-rule: a rule written before the trade that removes a decision you would otherwise make under pressure.
Which cognitive biases affect traders most?
Six biases are enough to explain most of the decisions that depart from a written plan: confirmation bias, loss aversion with the disposition effect, recency bias, overconfidence, anchoring and sunk cost. Each has original research behind it, and your own record is the evidence that matters for you.
| Bias | What the research found | Trading example | Written-rule counter |
|---|---|---|---|
| Confirmation bias | Nickerson (1998) reviews evidence that people seek and interpret evidence in ways that favour a belief they already hold. | After buying GBPUSD you read only bullish commentary, and skip the chart that argues against you. | Before entry, write one sentence on what would prove the trade wrong, and read it before touching a stop. |
| Loss aversion and the disposition effect | Kahneman and Tversky (1979): losses loom larger than equal gains. Odean (1998): investors realized 14.8% of paper gains but 9.8% of paper losses. | You close a winner at +0.6R for fear of giving it back, and let a loser run past your stop in the hope it recovers. | Set stop and target at entry. A stop may be moved to reduce risk, never to increase it. |
| Recency bias | Greenwood and Shleifer (2014): investor expectations of returns were strongly tied to recent stock returns. | After three wins you raise size from 0.10 to 0.40 lot. | Position size comes from a fixed risk percentage, and changes only at a scheduled review. |
| Overconfidence | Barber and Odean (2000): households that traded most earned 11.4% a year, while the market returned 17.9%. | A run of correct calls leads to more trades per day, more entries away from your setup. | Cap the number of trades per day and week, and log every trade, including the ones you skipped. |
| Anchoring | Tversky and Kahneman (1974): arbitrary starting numbers shifted people's estimates. | You refuse to close until the price "gets back to my entry". | Set exits from the setup and your risk limit, never from your entry price. |
| Sunk cost | Arkes and Blumer (1985): people continue an endeavour once money, effort or time is invested. | "I have held this for six weeks, so I will not close now." | Ask: would I open this trade now, at this price? If not, the plan decides, not the history. |
What is confirmation bias in trading?
Confirmation bias in trading is looking for, or reading, evidence in a way that supports a position you already hold. Nickerson's 1998 review in Review of General Psychology defines the term in that way and collects evidence of it across many kinds of reasoning.
Once you hold a position, you tend to notice supporting news more and give it more weight, so the trade feels better supported with no new evidence behind it.
The counter-rule is to write the opposing case before you enter. Add a level where you would act on it, and if that level is reached the note decides, not your mood.
What are loss aversion and the disposition effect?
Loss aversion is the finding that a loss hurts more than an equal gain pleases, and the disposition effect is the trading pattern it produces: selling winners too soon and holding losers too long. Kahneman and Tversky's 1979 prospect theory paper describes the value function as generally steeper for losses than for gains.
Odean (1998) tested the pattern on trading records for 10,000 accounts at a large US discount brokerage from 1987 to 1993. Investors realized 14.8% of their paper gains but only 9.8% of their paper losses. The paper found no sign that the behaviour was justified by later performance. It also found the pattern reversed in December, which the author links to selling losers for tax reasons. The study covered stock investors; your own journal shows whether the same pattern is in your trades.
The counter-rule is to fix the stop and the target when you enter. The common version is that a stop can be moved only to reduce risk, never to give the trade more room. A stop loss and take profit set before entry takes the decision away from the moment when loss aversion is strongest.
What is recency bias in trading?
Recency bias in trading is treating your latest results, or the market's latest move, as more informative than they are. Greenwood and Shleifer (2014) analysed six sources of investor expectations between 1963 and 2011 and found that expectations of stock returns were highly positively correlated with past stock returns, yet negatively correlated with model-based expected returns.
The typical signature is size. Three wins feel like proof of skill, so the next trade is larger; three losses feel like proof the method has stopped working, so the rules are dropped.
The counter-rule is a position size set by a fixed risk percentage. Size follows from the position sizing rule, not from the last three trades. It changes only at a scheduled review.
What is overconfidence in trading?
Overconfidence in trading is believing your judgement is more accurate than your record shows, and it tends to show up as too many trades. Barber and Odean (2000) studied 66,465 US households from 1991 to 1996, comparing how often each traded with what it earned.
The households that traded most earned 11.4% a year, while the market returned 17.9%, and the average household earned 16.4%. The authors argue that overconfidence can explain the high trading and the poor net results. Your own trade count, set against your results, shows whether the pattern applies to you.
A run of right calls can lead to entries away from your setup, and each extra trade also pays spread, commission and swap.
The counter-rule is a cap and a log. A maximum number of trades per day and week, and a journal entry for every trade, including the trades you decided not to take. Compare the confidence you wrote down before each trade with what happened.
What is anchoring in trading?
Anchoring is letting a number you were shown, or chose, pull your later judgement towards it. Tversky and Kahneman (1974) spun a wheel of fortune to get a number between 0 and 100, then asked people to estimate the percentage of African countries in the United Nations.
The median estimate was 25 for the group given a starting number of 10 and 45 for the group given 65. Payoffs for accuracy did not remove the effect.
In trading, the most common anchor is your own entry price. The market has never seen your entry, yet "I will close when it is back to my entry" ties the exit to it.
The counter-rule is to set exits from the setup and your risk limit. Ask where the idea is wrong, and how much you are willing to lose if it is. The answer to both comes from the chart and the plan, not from where you got in.
What is the sunk-cost fallacy in trading?
The sunk-cost fallacy is continuing with something because of what you have already put into it, even though that spending cannot be recovered by continuing. Arkes and Blumer (1985) found that theatre customers who paid more for a season subscription attended more plays over the next six months, and their questionnaire studies pointed to a wish not to appear wasteful.
In a trade, it sounds like "I have held this for six weeks", "I have already added twice" or "I cannot close at a loss now". None of those says where price is likely to go next.
The counter-rule is the clean-slate test. Ask: if I had no position, would I open this trade now, at this price, with this stop? If the answer is no, the plan is to close it. A written rule can also ban adding to a losing position, and adding money to an account to keep a trade open is never part of a plan.
How much can biases cost over ten trades?
Small distortions on each trade compound into a large gap between the plan and the result. This synthetic example assumes a $5,000 demo account, a risk of $50 per trade (1% or 1R) and a plan with a 1:2 risk-to-reward ratio, so a full win is +$100 (+2R). Ten trades are taken, and the same four win in both versions.
Version A: the plan is followed.
- 4 wins × $100 = +$400.
- 6 losses × $50 = −$300.
- Net: +$400 − $300 = +$100, which is +2R.
Version B: the biases act. Winners are closed early at +$30 (+0.6R), out of loss aversion. Losers are left to run to −$90 (−1.8R), because the stop was moved in hope of recovery.
- 4 wins × $30 = +$120.
- 6 losses × $90 = −$540.
- Net: +$120 − $540 = −$420, which is −8.4R and 8.4% of the account.
The gap between the two versions is $100 − (−$420) = $520. The win rate is 40% in both. Only the handling of the trades changed.
The example is simplified: moving a stop would also change which trades win. Its point is that an unrecorded habit and a recorded rule can produce very different results from identical entries.
Now the recency example in numbers. On EURUSD a pip on 0.10 lot is worth $1, so a 30-pip stop risks $30, or 0.6% of $5,000. After three wins the trader moves to 0.40 lot, where a pip is $4 and the same stop risks $120, or 2.4%. Risk per trade is four times larger, with no new evidence about the method.
How do skilled traders put bias research to work?
Skilled traders turn each bias into a rule written before the trade and a column in the journal, because naming a bias does not change a decision on its own. The research describes averages across many people. Your own record shows which of the six biases appear in your trades, and the rule is what changes the next order.
- Match the bias to the record: the brokerage studies covered US stock investors, and the other studies used experiments and questionnaires. Count the pattern in your own journal, such as stops moved or size raised after wins.
- Rules remove decisions: a counter-rule takes the choice away from the moment of pressure, so the trade is handled the way you planned it.
For the other half of the story, how to judge a decision when the outcome was good or bad by luck, see process over outcome.
Try it on demo
On a MetaTrader 5 demo account, take your last five demo trades from your trading journal and check three things. Did a stop ever move away from the price? Did any winner close well short of its target? Did your size change after a win or a loss? Write one counter-rule from the table into your written trading rules, then follow it for the next five demo trades.
Quick check
- You only read analysis that supports the position you already hold. This is
- In the Odean study, investors realized 14.8% of their paper gains but 9.8% of their paper losses. This pattern is called
- A trader raises size from 0.10 to 0.40 lot after three wins, with a 30-pip stop on EURUSD. Risk per trade goes from
- Which rule counters anchoring on your entry price?
- You have held a losing trade for six weeks and added to it twice. The clean-slate test asks
Frequently asked questions
What is confirmation bias in trading?
Confirmation bias in trading is seeking or reading evidence in a way that supports the position you already hold. It shows up as ignoring the chart or news that argues against you. Writing down what would prove the trade wrong, before entry, is a rule that counters it.
What is recency bias in trading?
Recency bias is treating your latest results or the market's latest move as more informative than they are. A typical sign is raising your position size after a few wins. A fixed risk percentage per trade, changed only at a scheduled review, counters it.
What is the disposition effect?
The disposition effect is the tendency to sell winning investments too soon and hold losing ones too long. In Odean's 1998 study of 10,000 brokerage accounts, investors realized 14.8% of paper gains but 9.8% of paper losses. A stop and target set at entry counter it.
Can you remove cognitive biases from your trading?
You reduce their effect rather than remove them, because they are normal features of human judgement. Rules written before the trade, a cap on trades per day and a regular journal review keep them away from your orders, and the journal shows which bias to work on next.
Sources
- Kahneman, D. and Tversky, A., Econometrica, Prospect Theory: An Analysis of Decision under Risk, accessed 19 October 2026: the value function is generally steeper for losses than for gains, and losses loom larger than gains. https://doi.org/10.2307/1914185
- Odean, T., The Journal of Finance, Are Investors Reluctant to Realize Their Losses?, accessed 19 October 2026: trading records for 10,000 brokerage accounts, 1987 to 1993, showed a proportion of gains realized of 0.148 against 0.098 for losses, and a reversal in December. https://faculty.haas.berkeley.edu/odean/papers%20current%20versions/areinvestorsreluctant.pdf
- Barber, B. M. and Odean, T., The Journal of Finance, Trading Is Hazardous to Your Wealth, accessed 19 October 2026: of 66,465 households, 1991 to 1996, the most active earned 11.4% a year against 17.9% for the market, and the average household earned 16.4%; the authors argue overconfidence can explain the trading level. https://faculty.haas.berkeley.edu/odean/papers%20current%20versions/individual_investor_performance_final.pdf
- Nickerson, R. S., Review of General Psychology, Confirmation Bias: A Ubiquitous Phenomenon in Many Guises, accessed 19 October 2026: confirmation bias is the seeking or interpreting of evidence in ways that are partial to existing beliefs, expectations or a hypothesis in hand. https://doi.org/10.1037/1089-2680.2.2.175
- Greenwood, R. and Shleifer, A., NBER Working Paper 18686 (published in Review of Financial Studies, 2014), Expectations of Returns and Expected Returns, accessed 19 October 2026: six measures of investor expectations, 1963 to 2011, were highly positively correlated with past stock returns and strongly negatively correlated with model-based expected returns. https://nber.org/system/files/working_papers/w18686/w18686.pdf
- Tversky, A. and Kahneman, D., Science, Judgment under Uncertainty: Heuristics and Biases, accessed 19 October 2026: median estimates of the share of African countries in the United Nations were 25 and 45 for groups given starting numbers of 10 and 65; payoffs for accuracy did not reduce the effect. https://doi.org/10.1126/science.185.4157.1124
- Arkes, H. R. and Blumer, C., Organizational Behavior and Human Decision Processes, The Psychology of Sunk Cost, accessed 19 October 2026: theatre customers who paid more for a season subscription attended more plays in the following six months. https://doi.org/10.1016/0749-5978(85)90049-4
Practise what you learned
Open an account to use MetaTrader 5. Start on a demo account before risking real money.
Open Account