Cognitive Bias During a Trading Drawdown

A trading drawdown can distort decision-making when a trader starts evaluating sizing, holding, and re-entry choices against recovering the prior equity peak, rather than against each setup’s own pre-existing criteria — the account’s distance from its high-water mark does the justifying work instead of the setup itself. Prospect theory and the break-even-effect research below supply the theoretical and experimental basis for that mechanism: risk attitude is reference-dependent, and framing a choice as a chance to get back to even after a loss has been shown, in laboratory choice tasks, to increase risk-seeking.¹² Neither literature studies discretionary traders or live brokerage accounts directly, so treat the prior equity peak becoming a trader’s operative reference point during a drawdown as a theoretically grounded application of that research to trading, not as a directly observed empirical finding about trading behavior.

This article’s scope is that state-driven mechanism: what can change while a decline stays open, as a sustained condition — not what one prior trade does to the next setup’s perceived quality, which is the narrower case recency bias after a trading loss covers. What a trading drawdown is defines the state itself; this article does not redefine it, and it does not address whether a losing outcome retroactively looks like a process failure during review — how outcome bias distorts decisions during a trading drawdown covers that separate, backward-looking diagnostic question.

Why a sustained decline moves the reference point, not just the balance

Prospect theory’s central finding is that risk attitude is not fixed — it flips around a reference point. People are typically risk-averse when a choice is framed as a gain relative to that point, and risk-seeking when the same choice is framed as a loss relative to it.¹ If a trader’s operative reference point during an open drawdown is the prior peak rather than the account’s current level, the account sits in the loss domain relative to that peak, and that framing alone — before any specific setup is evaluated — can predispose risk-seeking choices aimed at returning to the peak, a pattern that would look different if the same choice were evaluated against a neutral or gain-framed reference point instead.

This matters because a drawdown is a state, not an event: it persists across many decisions until equity returns to the reference peak, so a peak-anchored reference point would not reset trade to trade the way a single loss’s effect on the next setup can. A trader can requalify one next setup cleanly — passing the recency-bias check described elsewhere — and still be sizing every decision in that sequence against a peak-anchored reference point that a single-trade check would not catch.

The break-even effect: why “getting back to peak” increases risk-seeking

A more specific, choice-based finding sharpens the mechanism. Research on sequential gambling choices found that after a loss, subjects showed increased willingness to accept an otherwise unattractive bet when it was framed as a chance to get back to even — an effect distinct from loss aversion generally, since the same bet framed without reference to breaking even did not produce the same increased risk-seeking.² That research used laboratory gambling tasks with monetary stakes, not live trading accounts, and it does not by itself establish that a trader mid-drawdown reasons the same way about a specific setup — but it directly supports the claim this article makes about mechanism: risk-seeking increases specifically when a choice is framed against the prior reference point, not merely because a loss occurred.

Applied to a drawdown, one plausible operational failure mode is a position sized, or a setup accepted, because it offers a path back to the peak — a larger stop, a wider target, an instrument or setup outside the trader’s normal criteria — rather than because it independently qualifies. If that is happening, the peak is functioning as a magnet for the decision rather than remaining an inert historical fact for review.

Realized and open losses may not move risk-taking the same way

A drawdown can consist of realized losses, unrealized (“paper”) losses, or both, and the distinction is not cosmetic. Experimental research comparing the two found that, after a realized loss — the position closed and the loss booked — participants took on relatively less risk in a subsequent choice, while after a comparable but still-open, unrealized loss, participants took on relatively more risk in that same subsequent choice.³ One interpretation offered for the pattern is that an unrealized loss can stay psychologically bundled with the next risky decision, as though the original position and the new choice are still part of one open account, while realizing the loss closes that mental account and resets the reference point — a change the research associates with more conservative subsequent choices, not more aggressive ones.

That research used controlled experimental risk-taking tasks, not discretionary intraday trading or live brokerage accounts, and it does not establish that a trader’s account behaves the same way. This article does not claim that realizing a trading loss makes a trader more aggressive — the cited experimental pattern points the other way, toward reduced risk-taking once a loss is booked, and even that is evidence from a laboratory setting rather than a demonstrated trading rule. The practical implication is narrower: whether a given drawdown is built mainly from open, unclosed positions or from losses already realized may be relevant context for reviewing decisions made during it, not a fact that predicts which way any individual trader will lean.

Named biases most affected while a drawdown is open

Pattern	What changes while the state is open	Distinguish from	Where it’s owned
Loss aversion, escalated	Reluctance to realize a loss intensifies while the state stays open, since accepting the loss means accepting an outcome below the pre-drawdown reference point	Ordinary loss aversion on a single open trade	Loss aversion in trading
Disposition-style holding	A losing position is held past its own exit rule — continued exposure in the hope of recovery, rather than accepting an outcome below the reference point	A single trade’s exit management	Loss aversion in trading
Break-even-seeking / peak anchoring	Sizing or setup selection is justified by distance to the prior peak rather than the setup’s own criteria	Ordinary position sizing	Adjusting risk during a drawdown
Sustained recency extrapolation	”Nothing is working” is extrapolated across the whole stretch, not one trade, feeding avoidance or method abandonment	A single trade’s effect on the next setup	Recency bias after a trading loss
Loss-recovery re-entry	The next trade exists to repair the drawdown rather than because it qualifies	The peak-anchoring pattern above, which can occur without an explicit recovery motive	Revenge trading

Professional field evidence supports part of this table without proving all of it: a study of full-time futures floor traders found they held losing positions significantly longer than winning ones — the disposition effect — showing the pattern is not confined to inexperienced retail traders.⁴ That study did not find evidence that this specific holding-duration behavior itself was costly or reduced trading success, and it did not test the break-even-seeking mechanism described above directly. Treat it as evidence that asymmetric holding behavior is real in a professional population, not as proof that the behavior causes poor performance or as direct confirmation of the reference-point explanation above.

Observable signs the state, not the setup, is driving the decision

Position size or target distance justified by “how much is needed to get back to even,” rather than by the setup’s own risk unit.
Reluctance to exit a losing position that has already failed its own invalidation rule, specifically because exiting would lock in the loss.
A setup accepted outside normal criteria — a different instrument, timeframe, or size — because it offers a larger potential move toward the peak.
Reasoning, spoken or logged, that references the account’s distance from a prior high-water mark rather than the current setup’s conditions.
Resistance to a predefined risk step-down, experienced as the plan “punishing” the trader rather than as the plan working as designed.

None of these signs alone confirms the mechanism; a large position can also be independently justified by a high-conviction setup within existing risk rules. The distinguishing test is what the stated reason for the decision actually references.

A check before acting on a decision made mid-drawdown

Is this size or setup justified by its own criteria, or by the account’s distance from its prior peak? If the peak is doing any of the justifying work, the decision is reference-point-driven rather than setup-driven.
Would this decision look the same if the account were flat instead of in a drawdown, with the setup unchanged? If not, the drawdown state — not the setup — is producing the difference.
If closing a losing position is being avoided, is that because the setup remains valid, or because closing would realize the loss? The second reason is the pattern this article describes, not a trade management decision.
Is the reasoning invoking a rule that existed before the drawdown began, or is it inventing a justification after the fact? A predefined risk-ladder rule — covered in full in adjusting risk during a drawdown — is the state working as designed; an after-the-fact justification for taking on more risk is not.

These questions test the reasoning behind the decision, not whether the trade eventually works out. A trade can pass this check and still lose; a trade that fails this check can still occasionally win, which is exactly what makes the pattern hard to see without a review process.

Reviewing decisions made while a drawdown was open

A single instance does not establish a pattern, and reviewing only the decisions a trader remembers as “the bad ones” builds in the same distortion the review is supposed to catch. A cleaner accounting separates occurrence from classification:

eligible occasion =
  a sizing, entry, or exit decision made while account equity
  was below its reference peak

reviewable =
  the logged reason for the decision is specific enough to
  classify without relying on hindsight

classifiable =
  the reviewable reason can be sorted into peak-referenced,
  rule-referenced, or setup-referenced

peak-referenced =
  the stated reason invokes the prior peak, "getting back,"
  or avoiding realization of an open loss

rule-referenced =
  the decision follows a risk-ladder or re-entry rule defined
  before the drawdown began

unclassified =
  the occasion occurred, but the logged reason is missing
  or too vague to classify

Report a rate only across classifiable occasions, and report the unclassified share alongside it rather than treating missing evidence as clean:

peak-reference rate =
  peak-referenced occasions / classifiable occasions

unclassified share =
  unclassified occasions / eligible occasions

A high peak-reference rate is evidence of a recurring pattern worth addressing through the risk-ladder and re-entry rules covered elsewhere — it is not, by itself, proof that any single decision it counted was wrong, and a low rate does not certify that every decision in the drawdown was well-reasoned; it only means the logged reasons did not reference the peak. This method classifies logged reasoning, not a trader’s cognitive state: it does not diagnose bias with certainty, and it does not determine whether any individual decision was objectively profitable or wrong.

Where this connects

This article covers the interpretation layer: how being in a drawdown can change what a decision is justified by. It does not cover measuring the drawdown itself — see what a trading drawdown is — building the risk-reduction ladder that should apply once the state is recognized — see adjusting risk during a drawdown — or the specific pattern of a trade organized around recovering a loss, which revenge trading owns directly. Loss aversion in trading and recency bias after a trading loss cover the two named biases most likely to intensify while a drawdown is open, at the single-trade level this article extends to the full stretch.

Where Costante fits

Costante supports logging the stated reason for a sizing, entry, or exit decision alongside the account’s drawdown state at the time, so a peak-referenced justification is visible in review rather than reconstructed from memory afterward. Predefined risk-ladder states and re-entry rules can be recorded before a drawdown begins and stay visible at the decision moment, giving the check above something concrete to compare against.

Costante does not calculate or monitor drawdown from a connected account, does not detect cognitive bias with certainty, does not decide whether a specific decision was justified, and does not automatically block or resize a trade. The trader defines the risk rules, logs the reasoning, and remains responsible for every decision.

Frequently asked questions

Is this the same as revenge trading?

No. Revenge trading is a specific action pattern: a trade organized around recovering a loss. The mechanism this article describes is broader and can occur without an explicit recovery motive — a trade can be peak-anchored in its sizing or setup selection without the trader consciously framing it as “getting back” at anything. The two frequently co-occur, but a decision can show one without the other.

Does this only apply to open, unrealized drawdowns?

No, and the direction is easy to get backward. Experimental evidence comparing realized and unrealized losses found greater subsequent risk-taking after paper (unrealized) losses than after losses that had been realized in the studied setting.³ That evidence comes from controlled experimental tasks, not discretionary intraday trading, and it does not establish the same magnitude or mechanism holds for a live account. Treat realized-versus-unrealized status as potentially relevant review context — a reason to log which kind of loss is driving a given drawdown — rather than as a deterministic rule for how any specific drawdown will play out.

Does a shallow drawdown produce the same effect as a deep one?

The cited research does not establish a depth threshold, and this article does not introduce one. The proposed mechanism concerns being below a reference peak at all, not a specific depth — but a shallow, brief decline gives fewer decisions time to accumulate a peak-referenced pattern than a deep or prolonged one does, which is a practical reason deeper or longer drawdowns warrant closer review, not evidence that shallow ones are exempt from the mechanism.

Once equity returns to the prior peak, does the distortion end immediately?

Once equity reaches the prior peak, the specific condition this article defines as an open drawdown is no longer present, so “distance to the prior peak” no longer describes the same state. What a trading drawdown is covers when the equity measurement itself returns to zero. That does not establish that any decision habit, emotional response, or behavioral pattern that developed while the drawdown was open disappears immediately — reviewing those decisions is a separate step, addressed by the review method above, not something the equity measurement resolves on its own.

How is this different from recency bias after a single loss?

Recency bias after a trading loss covers how one prior trade’s outcome distorts the very next setup’s perceived quality. This article covers a different, state-level mechanism that can persist across many trades for as long as the account remains below its reference peak, independent of what any individual prior trade was. A trader can pass the single-trade recency check on every setup in a sequence and still be size-anchoring every one of those setups to the account’s distance from its peak, if that peak has become the operative reference point for those decisions.

Sources

Costante provides educational workflow tools, not financial advice. Trading involves risk.

Kahneman, D., & Tversky, A. (1979). Prospect Theory: An Analysis of Decision under Risk. Econometrica, 47(2), 263–291. ↩ ↩²
Thaler, R. H., & Johnson, E. J. (1990). Gambling with the House Money and Trying to Break Even: The Effects of Prior Outcomes on Risky Choice. Management Science, 36(6), 643–660. ↩ ↩²
Imas, A. (2016). The Realization Effect: Risk-Taking after Realized versus Paper Losses. American Economic Review, 106(8), 2086–2109. ↩ ↩²
Locke, P. R., & Mann, S. C. (2005). Professional Trader Discipline and Trade Disposition. Journal of Financial Economics, 76(2), 401–444. ↩