Sunk Cost Fallacy in Trading: What the Evidence Supports and What It Cannot Explain

By Sofia Harchich | Trading Psychologist & Behavioural Finance Writer | thewealthmirror.com

The sunk cost fallacy in trading is the most frequently cited explanation for a position held too long, and it explains rather less of that behaviour than its popularity suggests.

The definition is sound. The sunk cost fallacy is the tendency to continue with something because of what has already been invested in it — time, money, effort, emotional energy — when abandoning it is clearly the better course. Applied to a trade: the position is held because of what was already put in, rather than because of what it is now likely to do.

The trouble comes when the sunk cost fallacy in trading gets stretched to cover behaviours it was never about. Reading it carefully produces a sharper tool and rules out a couple of interventions that waste months.

Finding one: the effect is real and it is about consistency, not money

Supported: strongly, across many domains.

The classic demonstrations have nothing to do with markets. People finish meals they are no longer enjoying, sit through films they dislike, and continue projects they know will fail — reliably, and in proportion to what has already gone in.

The mechanism most consistently supported is a preference for consistency with past commitments rather than financial optimisation gone wrong, plus a reluctance to have wasted something. What is being protected is the coherence of a sequence of decisions, and money is only one of the things that can be sunk into it.

For trading this is the first useful correction. The sunk cost fallacy in trading is not primarily about the money already lost on the position. It is about the analysis, the waiting, the conviction expressed to yourself at entry — the things that would be retroactively wasted if the position were closed now.

Finding two: prior effort weighs more heavily than prior money

Supported: moderately well, and it changes which trades are at risk.

Across studies, effort invested tends to produce stronger sunk cost effects than money invested at equivalent nominal value. Time spent, work done, and attention given attach more firmly than a comparable sum simply spent.

The trading implication is specific and testable on your own record. The positions most vulnerable are the ones you worked hardest on rather than the largest ones — the thesis you built over a weekend, the setup you waited three weeks for, the instrument you have been studying since spring.

That predicts something a purely financial account does not: a small position built on a great deal of research should be held longer past invalidation than a large position taken on a routine signal. Most traders assume the opposite. The record usually agrees with the research.

The position that is hardest to close is rarely the biggest one. It is the one you did the most work on.

Finding three: the effect weakens when the future is stated separately

Supported: reasonably well, and this is where the intervention comes from.

The most reliable experimental way to reduce sunk cost effects is to make the forward-looking question explicit and separate from the history — asking what the best course is from here, given only what lies ahead.

This works because the fallacy operates by fusing two questions that are logically independent: what have I put in and what is this likely to do next. Separating them removes most of the pull without requiring anyone to care less about what they invested.

In practice this is a written question, asked at a scheduled moment rather than when the position feels uncomfortable: if I had no position here and no history with this idea, would I enter now, at this price, at this size? A no is a close. The question is trivially easy at 09:00 and nearly impossible to ask spontaneously at 14:30, which is why it belongs on a timer rather than in your judgement.

Finding four: it compounds with loss aversion rather than adding to it

Supported: as a straightforward implication of both literatures.

Loss aversion makes the realised loss hurt roughly twice as much as an equivalent gain feels good. Sunk cost makes the prior investment count as a reason to continue. Together they answer each other’s objections.

Confronted with this position is unlikely to work from here, sunk cost supplies but look at what has gone into it. Confronted with then close it, loss aversion supplies not yet, closing makes it final. Neither argument survives alone for long. In combination they can hold a position indefinitely, and the trader experiences this as careful deliberation rather than as two biases taking turns. Held long enough, the pair can turn a hold into an add, which is the sequence reconstructed in averaging down a losing trade.

What the sunk cost fallacy in trading cannot explain

Here is the part that matters, and the reason a careful reading is worth more than a definition.

It does not explain re-entry. A trader who closes a losing position and re-enters twenty minutes later at double size has not committed a sunk cost error — the cost was realised, the commitment was broken, and the sequence was abandoned. That is a different mechanism entirely, and it belongs with how to stop revenge trading.

It does not explain the timing. Sunk cost predicts that a position will be held. It says nothing about why the trader who was reluctant at 10:20 finally acts at 15:40, or why the delay is longer on some instruments than others. The effect is stable across a decision; the behaviour it is invoked to explain varies hour to hour.

It does not explain the losses on positions with no prior investment. A significant proportion of over-held trades are routine entries with no thesis, no research and no waiting — nothing sunk except the entry itself. Whatever holds those open is not this.

And it does not explain why knowing about it changes nothing. Sunk cost is among the most widely known biases in retail trading. Awareness of it is close to universal and holding times have not improved, which is a strong indication that the mechanism is not primarily a reasoning error at the point of decision.

That last one deserves the weight. A bias that persists undiminished in people who can define it accurately is being maintained by something other than the misunderstanding it names.

Of the five steps in the instability loop, the sunk cost effect sits closest to the belief at the end: what has already been spent becomes a reason to stay consistent, and consistency then decides the next behaviour.

The stage where it actually lives

Placing the sunk cost fallacy in trading inside the loop resolves the puzzle.

The circuit runs: trigger, activation, behaviour, outcome, reinforced belief. Sunk cost is usually taught as though it operated at the behaviour stage — a faulty calculation made while deciding. Its real work is at the reinforced-belief stage, one cycle earlier.

Each time a position is held and eventually recovers, the belief waiting works is strengthened, and the threshold for holding drops slightly. Each time it does not recover, the belief I closed too early last time is available instead, because the counterfactual is unobservable. Both outcomes feed the same conviction, which is why the pattern is immune to experience in a way that a simple calculation error would not be.

This is also why the effective interventions are the ones that operate before the position exists. A decision made at entry is made outside the circuit. A decision made at 14:30 is made inside it, by a trader whose belief about waiting has been reinforced by every previous instance regardless of how those instances turned out.

What to do with the four findings

1. Log research time per trade, not just size. Finding two says this column predicts your holding behaviour better than the size column does
2. Write the forward-looking question into your routine, at a fixed time rather than at a feeling: would you enter this now, flat, at this price
3. Treat the compound as one problem. Interventions aimed at sunk cost alone leave loss aversion supplying the same conclusion by a different route
4. Stop attributing every over-held position to it. Three of the four common cases above are something else, and the something else has its own remedy

Testing it on your own record in two weeks

The sunk cost fallacy in trading is checkable against your own trades, and the test is more informative than any amount of reflection on whether the sunk cost fallacy in trading applies to you.

Add two columns to your log for twenty trades. Research time — minutes or hours spent on the idea before entry, roughly. Minutes past invalidation — how long the position stayed open after the condition that should have closed it occurred.

Then plot the second against the first, and separately against position size.

Three outcomes are possible, and each points somewhere different.

Minutes past invalidation tracks research time. Finding two is running. The remedy is the scheduled forward-looking question, applied specifically to your most-researched positions rather than to all of them, because those are the ones where the effect concentrates.

It tracks position size instead. This is not sunk cost. Size-driven delay is loss aversion operating on the magnitude of the realised loss, and it responds to changing the reference point rather than to any question about prior investment.

It tracks neither. The delay is being produced by something outside both mechanisms — most often state, in which case the pattern will correlate with the pre-session word rather than with anything about the trade.

Two weeks of data settles which of the three you are dealing with, and the distinction is worth more than a fortnight of reading. Traders routinely spend months applying a sunk cost remedy to what turns out to be the second case, conclude the remedy does not work for them, and add it to the list of things that failed.

One practical note on the research-time column. It only works if logged at entry rather than reconstructed later, because the estimate afterwards is coloured by the outcome — a position that worked gets remembered as well-researched, and one that failed gets remembered as hasty. Behavioural anchoring applies to your own log as much as to anything else.

The part that is not a fallacy

One closing observation, because the word fallacy does some quiet damage here.

The reluctance to waste work is not a defect in reasoning. In nearly every domain outside markets, persisting with something you have invested heavily in is exactly right — it is how skills get acquired, relationships survive difficult years, and long projects reach completion. The tendency is doing something valuable almost everywhere, and markets are one of the few environments where the past investment genuinely carries no information about the future.

Which means the sunk cost fallacy in trading is less a flaw you carry than a well-adapted instinct meeting an unusual environment. That reframe is offered as mechanism rather than as consolation, and it is why the fix is a written question at a scheduled time rather than an attempt to become the sort of person who abandons things easily — because you would not want to be that person anywhere else.

For the wider sorting framework, start with behavioural finance for traders; for the compound with identity, loss aversion, ego and the trading account. The pattern quiz identifies which emotional pattern your behaviour under pressure fits.

---

About the Author

Sofia Harchich is a Trading Psychologist and Behavioural Finance Writer with a Master’s in Psychology. She works at the intersection of Jungian shadow work, neuroscience, and market behaviour — helping traders understand the psychology driving their decisions, not just the strategy.

Read more at thewealthmirror.com/about