The biases that cost the most

Everything you've read so far works in a spreadsheet. The problem shows up when there's real money at stake and a person is deciding. These five patterns are documented across real accounts, they happen to almost everyone, and knowing their names isn't enough to avoid them.

Why you sell what's rising and hold what's falling

It's the best-measured pattern of all and it has its own name: the disposition effect. It consists of closing positions doing well and keeping those doing badly, which is precisely the opposite of what the previous chapter's arithmetic recommends.

The measurement

Terrance Odean, of the University of California, Berkeley, analysed 10,000 accounts at a large US discount broker between 1987 and 1993, publishing the result in 1998 in The Journal of Finance.

He measured two proportions in each account:

• Of all the gains an investor could realise at any given moment, they realised 14.8%.
• Of all the losses they could realise, only 9.8%.

The ratio between the two exceeds 1.5. In plain terms: a stock showing a gain is more than 50% more likely to be sold on any given day than one showing a loss.

Odean also checked that the behaviour wasn't explained by wanting to rebalance, nor by avoiding the costs of trading low-priced shares. And subsequent performance didn't justify it: the stocks sold kept rising and those held kept falling.

Why it happens

Because closing a losing position requires doing something deeply uncomfortable: admitting in writing that the previous decision was bad. While the position stays open, the loss is a reversible possibility. The moment you sell, it's a fact.

Selling a gain does the opposite: it turns a possibility into a verified win, and that feels excellent.

The result is the exact pattern from the previous chapter's first session: many small gains and a few enormous losses, with a high hit rate and a falling account.

What can be done

Naming it isn't enough. What works is removing the decision from the moment the bias operates.

If the exit point is written before buying — as the position sizing session set out — closing a loss stops being a decision made under pressure and becomes the execution of a decision already made. It doesn't remove the discomfort, but it moves it to a moment where it costs nothing.

There's a detail from the study worth recalling here: Odean observed that tax-motivated selling concentrated in December. Meaning people do realise losses when they have an external reason justifying it. A written rule performs exactly that function for the rest of the year.

Why being right feels like skill and being wrong like bad luck

It's the combination of two things: attributing wins to your own ability and losses to external factors, and from there believing you're better than you are.

The mechanism, in two sentences you'll recognise

"I bought well, I could see the sector was going to move."

"It was going fine until that news came out that nobody could have predicted."

Both can be true in a specific case. The problem is when the pattern always runs in the same direction, because then it isn't describing what happened: it's protecting the image someone holds of themselves.

Where it leads

To the practical consequence that names the next session. If every win confirms you know what you're doing and every loss was bad luck, the inevitable conclusion is that you should trade more, because your judgement works and the losses weren't attributable to it.

That chain — attributing wins inward, losses outward, concluding you should trade more — is the explanation the researchers themselves give for the phenomenon we'll see next session with numbers.

The close cousin worth knowing

There's a third pattern acting before those two that almost nobody spots in themselves: seeking information confirming what you already think and dismissing what contradicts it.

In practice it works like this. Someone buys a stock, and from that moment the favourable articles look well argued and the unfavourable ones look alarmist. They aren't lying: they're filtering without noticing.

The check that detects it is one question: when was the last time I read something that changed my mind about a position I currently hold? If the answer is "never," the filter is running at full capacity.

How the first number you saw anchors your decision

Anchoring means a number you saw earlier conditions your later estimate, even when that number bears no relation to what you're deciding.

The experiment that demonstrated it

Amos Tversky and Daniel Kahneman published it in 1974 in Science. They spun a rigged wheel of fortune in front of participants and then asked for an estimate on a question with nothing to do with the number that came up.

The estimates shifted toward the wheel's number. An irrelevant figure, generated in plain sight by a random mechanism, moved the answer.

How it shows up with money

The price you paid. It's the most powerful anchor there is, and it's completely irrelevant to the future. The market doesn't know what you paid, and the company couldn't care less. Yet that figure decides, for many people, when to sell.

The all-time high. "It was worth $80, now it's at $45, it's cheap." That $80 is a price that existed under conditions that may no longer exist. The session on multiples already explained why that alone says nothing.

The first price target you read. An analyst publishes an estimate and the number sticks, even after five other ones come out.

The test that disarms it

One question, and it works because it forces reasoning without the anchor: if I didn't hold this position and saw this company today for the first time, would I buy it at this price?

If the answer is no, the only thing holding you is the number you entered at. And that number isn't part of the company's business.

Why losing hurts twice as much as winning

Losing a hundred euros produces markedly more distress than gaining a hundred produces pleasure. It isn't an impression: it's a central finding of the prospect theory Daniel Kahneman and Amos Tversky published in 1979, and it explains much of what we've seen so far.

What it implies in practice

If the pain of losing outweighs the pleasure of gaining, two behaviours become predictable.

Holding a losing position, because closing it crystallises a pain that can be postponed. It's the disposition effect again, seen from its emotional cause.

Closing a winning one too early, because securing the pleasure feels more urgent than maximising it.

Both behaviours together produce exactly the pattern Odean measured.

The twist worth understanding

Here's something prospect theory describes that runs counter to intuition: once someone is already in a loss, their willingness to take risk increases.

It's the mechanism of the gambler who's losing and raises the bet to win it all back at once. Applied to a portfolio, it's the person who after a bad run decides to concentrate more into a single idea to get back in the green sooner.

It's precisely the opposite of the anti-martingale principle from the position sizing session, which is why that session insisted on calculating risk against current capital: so size falls on its own exactly when instinct pushes to raise it.

And a consequence for the course itself

Loss aversion explains why the previous chapter's recovery arithmetic weighs so heavily. A 50% drawdown doesn't only require a 100% rise to undo: it requires staying invested throughout that process, with the pain weighing double each month that passes.

That's the real reason most portfolios needing a 100% rise never got it. It wasn't that the market didn't recover: it's that the person sold along the way.

Why money doesn't have compartments

Mental accounting means treating identical amounts of money differently depending on where they came from or which mental drawer they sit in. The term is Richard Thaler's, and it describes something real accounting doesn't permit.

The three cases that show up most

"House money." Someone makes €2,000 on a position and decides to risk it on something highly speculative, arguing "this is profit now, it isn't my money." It is. The moment the gain exists, those €2,000 are worth exactly what the €2,000 they started with are worth, and losing them hurts the same at tax time.

Sealed drawers. Someone keeps €5,000 in a 0% account while carrying €3,000 of card debt at 20%, because the first is "savings" and the second is "a separate matter." They aren't two matters: it's one financial situation where 20% is being paid for not using money that earns nothing.

The emotional position. Shares inherited from a relative, or shares in the company where you work, get treated under different rules from the rest of the portfolio. They're the same shares with the same risk, and precisely the ones least often reviewed.

The rule that cuts it

All your money is the same money. A euro made speculating, a euro inherited, and a euro from your salary are worth exactly the same and deserve the same criteria.

That doesn't mean you can't separate goals: Chapter 2's profile sheet did exactly that, and rightly, because house deposit money and retirement money have different horizons.

The difference is in the criterion. Separating by horizon is planning. Separating by origin, to justify taking more risk with one part, is mental accounting.

How the five link together

The biases don't act separately: they feed each other in a fairly predictable order, and seeing the whole chain explains more than seeing them individually.

The filter starts. You buy something and from then on the information confirming it looks solid and the information contradicting it looks overblown.

The anchor takes hold. The price you paid becomes the reference you measure everything against, even though the market doesn't know it.

Loss aversion acts. The position goes badly and closing it hurts more than it would relieve, so it gets postponed.

And the disposition effect appears. Gains get closed to secure the pleasure and losses get held to postpone the pain. It's the consequence of the previous three, not a separate cause.

The outcome gets misattributed. What went well confirms your own judgement; what went badly was circumstance. The conclusion is that you can trade more.

And mental accounting opens the door. Recent gains get treated as different money, so risking them feels less serious.

Back to the start, with more money at stake.

Why this matters more than the list. None of the five gets fixed by knowing about it, because all of them act before there's anything to think about. What does work is breaking the chain at the only point where you can: before entering. An exit point written before buying disarms the anchor, loss aversion, and the disposition effect all at once, because by the time the moment arrives the decision is already made.

That's what the two sessions closing this chapter are about.

What to remember

• Odean measured across 10,000 accounts that 14.8% of available gains get realised and only 9.8% of losses: a stock showing a gain is more than 50% more likely to be sold.
• Attributing wins to skill and losses to bad luck leads directly to trading more, which is the next session's subject.
• The price you paid is the most powerful anchor there is and it's completely irrelevant to what the company is worth.
• Once already in a loss, willingness to take risk increases: precisely when position size should be falling.

Published August 3, 2026. Last reviewed: August 3, 2026.

Related: why being right isn't the same as winning · how much to risk on any single trade