Loss aversion in bear markets
What these dates mean
Published is the original release. Updated records a material content change. Reviewed records the latest documented factual or editorial review; it does not mean personalized professional advice.
Losses hurt more than equivalent gains feel good — how much more is genuinely unsettled, and the honest version still tells you what to do.
Imagine you find a $100 bill on the sidewalk. You feel a quick burst of pleasure. Now imagine you notice a $100 bill has fallen out of your pocket. The irritation lasts longer and bites harder than the pleasure did.
That asymmetry is loss aversion, and it is the reason a falling market provokes action while a rising one does not. This spoke sits under the Money psychology hub.
How much more do losses hurt? Less precisely than you have been told
The usual line is "losses hurt exactly twice as much." That number comes from prospect theory's loss-aversion coefficient, and it is worth being careful about what it does and does not say.
The largest meta-analysis to date pooled 607 estimates from 150 articles and found a mean coefficient of about 1.96, with a 95% interval of 1.82 to 2.10 (Brown, Imai, Vieider & Camerer, 2024). So "about two" is a fair summary of the average.
Two caveats that most articles skip:
- The spread is wide, and poorly explained. Individual estimates scatter well beyond that interval, and the authors found that few observable features of study design account for the differences. An average of noisy estimates is not a law of nature.
- The magnitude is contested. A 2025 reanalysis of the same dataset (Yechiam & Zeif) argues loss aversion is not robust once you account for how the estimates were produced. That does not make the phenomenon vanish, but it does mean the confident "exactly 2×" you see quoted everywhere is over-precise.
And the stakes matter. These estimates come overwhelmingly from small-to-moderate laboratory bets. Nobody has established that a $10,000 loss is felt exactly twice as intensely as a $10,000 gain, and we are not going to be the site that pretends otherwise. What survives the extrapolation is the direction, not the multiple — and the direction is enough to explain the damage.
The part that did hold up: the disposition effect
Here is a finding in real money, from real brokerage accounts, that has been reproduced repeatedly. Odean (1998) examined 10,000 accounts and found investors were markedly more likely to sell their winners than their losers — even though the behaviour was not explained by tax, rebalancing or subsequent performance. Selling a winner realises a gain, which feels good. Selling a loser forces you to admit a loss, which does not. So the loser stays.
The same wiring explains the crash reflex. When a portfolio drops 20%, the pain is not abstract, and the mind offers one very appealing way to stop it: sell everything and go to cash.
Why selling is the actual mistake
A decline is a real decline. What it is not — yet — is permanent.
Selling during a crash converts a paper loss into a realised one and guarantees you miss the recovery, which historically arrives without warning and without asking whether you are back in. The market does not send a note. That is the whole mechanism by which loss aversion destroys wealth: not the feeling, but the trade the feeling recommends.
How often does this actually happen?
Often enough that you should expect it, and not on a schedule anyone can quote you. Be suspicious of any article — including an earlier version of this one — that tells you "the market drops 20% every 5 to 7 years." The count depends entirely on the definition and the era:
| How you count | Result |
|---|---|
| 20% peak-to-trough declines, S&P 500, since 1928 (Ned Davis definition) | 27 bear markets — an average of roughly one every 3.5 years |
| The same data, 1928–1945 only | 12 — roughly one every 1.5 years |
| The same data, 1945 to today | 15 — roughly one every 5 years |
| Stricter closing-basis definitions used elsewhere | As few as 13–15 since 1929 |
Same market, same century, answers that differ by a factor of two. The takeaway is not a frequency. It is that you should expect several in an investing lifetime and none of them will be announced.
If you want to see what a bad stretch actually does to a portfolio you are drawing from, the sequence-risk stress test shows it in numbers rather than adjectives — and sequence-of-returns risk is why the first decade of retirement is the dangerous one.
How to protect yourself
- Decide before, not during. Write down what you will do in a 30% drawdown while nothing is happening. The plan you write calmly is the only one worth having.
- Automate contributions. If the money moves on payday without your involvement, a crash automatically means buying more shares — with no act of courage required.
- Look on a schedule you chose. Check quarterly because you decided to, not hourly because the news told you to.
- Do not average the pain ratio into a plan. "Losses feel worse" is enough. You do not need a coefficient to know that selling at the bottom is the thing to avoid.
Key takeaways
- Losses do loom larger than gains. The average coefficient is about 2 — but the estimates vary widely, the magnitude is contested, and it is not established at large stakes.
- The finding that is solid in real markets is the disposition effect: investors sell winners and hold losers.
- Selling in a crash is what turns a decline into a permanent loss. The feeling is not the damage; the trade is.
- Bear-market "frequency" depends on the definition. Expect several in a lifetime; do not plan around a schedule.
Educational only — not financial advice.
FAQ
Do losses really hurt twice as much as gains feel good?
On average, roughly — but with less precision than you have been told. The largest meta-analysis (607 estimates) puts the average coefficient at about 1.96, with a 95% interval of 1.82 to 2.10. Individual estimates scatter widely, and a 2025 reanalysis of the same data argues the average is not robust.
Does loss aversion apply to a $10,000 loss the same way?
Nobody has established that. The estimates come overwhelmingly from small-to-moderate laboratory stakes. Extrapolating a lab ratio to a five-figure portfolio drawdown is a guess, and we would rather say so than give you a confident number. What survives extrapolation is the direction, not the multiple.
What part of loss aversion is actually well established in investing?
The disposition effect. Odean (1998) found investors sell their winners and hold their losers — behaviour consistent with an unwillingness to realise a loss. It has been reproduced in the US, Europe and Asia across decades. That is a much firmer footing than any specific pain ratio.
How do I avoid panic selling in a downturn?
Decide what you will do before the drop, not during it. A written plan and an automatic contribution schedule remove the moment of decision, which is exactly when loss aversion is loudest. Reducing how often you look also helps, and costs nothing.
How often do bear markets actually happen?
Depends entirely on the definition, which is why we will not give you one tidy number. Counting 20% peak-to-trough declines, one widely used dataset lists 27 since 1928 — but 12 of those fell between 1928 and 1945. Since 1945 the pace has been closer to one every five years. Plan for 'several in an investing lifetime', not a schedule.
Is a paper loss really not a loss?
It is a real decline in the value of what you own. The distinction that matters is that it is not yet *permanent*. Selling converts a decline that may reverse into one that cannot. That is the mechanism by which loss aversion actually destroys wealth.
Should I just stop looking at my portfolio?
Looking less during a decline is a defensible tactic — investors already do it involuntarily, which is its own problem. The useful version is deliberate: check on a schedule you set in advance, not in response to a headline. Avoidance you chose is a plan; avoidance that chose you is the ostrich effect.
Sources and notes
- Prospect Theory: An Analysis of Decision under RiskKahneman & Tversky, Econometrica 47(2), 1979 · Accessed 2026-07-11The original evidence that losses loom larger than equivalent gains, and the origin of the loss-aversion coefficient.
- Meta-analysis of Empirical Estimates of Loss AversionBrown, Imai, Vieider & Camerer, Journal of Economic Literature 62(2), 2024 · Accessed 2026-07-11607 estimates from 150 articles. Mean loss-aversion coefficient 1.955, 95% interval [1.820, 2.102]. Source of the 'about 2 on average' figure and of the point that few study features explain the spread.
- Loss Aversion is Not Robust: A Re-Meta-AnalysisYechiam & Zeif, Journal of Economic Psychology, 2025 · Accessed 2026-07-11Reanalysis of the same dataset arguing the ~2 average is not robust. Used for the honest caveat that the magnitude is contested.
- Are Investors Reluctant to Realize Their Losses?Odean, The Journal of Finance 53(5), 1998 · Accessed 2026-07-1110,000 brokerage accounts. Investors were markedly more likely to sell winners than losers — the disposition effect. Widely reproduced across markets; used as the finding that did hold up.
- 10 Things You Should Know About Bear MarketsHartford Funds (data: Ned Davis Research) · Accessed 2026-07-1127 bear markets in the S&P 500 since 1928 on their definition; 12 between 1928 and 1945, 15 since. Used for the frequency table and the point that the count depends entirely on the definition.