Sunk-Cost Fallacy — When to Sell Stocks Even at a Loss
“I can’t sell now, I’ve already lost $5,000.” That sentence is the most expensive in retail-investor vocabulary. What you paid is sunk: the money is gone, regardless of what you do next. The only relevant question is: would I buy this today at the current price? Here are six trigger questions, clear mechanics, and three real cases where selling at a loss was the only correct call.
What the sunk-cost fallacy is
Arkes/Blumer 1985 demonstrated the effect: subjects who paid for a theatre ticket attend the show even in bad weather. Subjects with a free ticket stay home much more often. The money already paid does not come back — regardless of action. But the human brain treats it as if a “paid” account had to be settled against a benefit.
Sell decision is purely a function of future expectation — orthogonal to cost basis. What you paid cannot move current capital. It can only move your perception — and that is the trap.
Six trigger questions justifying a loss-sale
- Would I buy this stock today at the current price? If no → sell. Doesn’t matter whether the loss is 10% or 60%.
- Has the thesis changed? New competitors, regulatory issues, management change, debt up. If yes → sell.
- Is there a better alternative for the same capital? If another position has 5%+ higher annual return expectation → switch.
- Do I have an emotional attachment to the stock? “My stock”, “my favourite sector”, “my first position” → sell, because emotional attachment amplifies the bias.
- Would a neutral advisor sell? Imagine you inherit exactly this portfolio. Would you keep it as-is? If no → sell, because in fact you do inherit it (every day, from yourself).
- Does the position affect your sleep? If you think about it at night, two possibilities: position size too big or thesis too weak. Either → sell or reduce.
Three real cases for the right sell at a loss
- Wirecard 2018 (special-audit phase): Investors at €100 cost basis saw the stock halve to €50 in 2018. Selling during the special-audit phase avoided the total loss two years later. “I can’t accept a 50% loss” became “I lost 100%”.
- Tilray 2019 (cannabis-mania top): Investors at $250 saw the stock fall to $120. Selling in Q1 2019 saved the fall to $1.80 today. The 50% loss looked like “it’ll come back” in 2019, but it was the top of the cannabis bubble.
- Bayer 2020 (glyphosate phase): Investors at €95 saw the stock fall to €50. Selling in the first glyphosate wave 2020 meant −47%. Holding meant −74% today. Holding cost a second halving.
In all three cases the rational decision was: sell, realise the loss, redeploy capital into better alternatives. The sunk-cost fallacy held many investors back and amplified the damage by 2-50x.
The most expensive variant: “averaging down to break-even”
An especially aggressive sunk-cost form: cost-averaging down on a falling stock. Logic: “Bought at $100, now at $50, average $75 — only needs to rise to $75 for break-even.”
If the thesis is intact, averaging down is rational. If the thesis is shaky, averaging down is loss amplification.
When holding is right
Sunk-cost fallacy doesn’t mean “always sell when red”. Holding is right when:
- Fundamental business metrics are stable or better than at purchase (cashflows, margins, market share)
- The price drop is market-driven (sector correction), not company-specific
- You would buy fresh today — not because cost basis was $100, but because $50 is cheap
- You have a 5-10 year horizon and can sit through volatility
- Position size is manageable (max 5-10% of portfolio) — even total loss doesn’t change the life plan
How much does the sunk-cost fallacy cost?
| Study | Finding | Return drag |
|---|---|---|
| Arkes/Blumer 1985 (original) | Sunk-cost subjects make worse forward decisions | Variable by setting |
| Statman/Caldwell 1987 (stocks) | Investors with loss positions hold 2.5x longer than rational | 1-2% p.a. drag |
| Coval/Shumway 2005 (traders) | Even pros show sunk-cost — measured intraday | 0.5% per day on affected trades |
| BMI investor data 2024 | 34% of held loss positions wouldn’t be bought today by the same investor | Structurally significant |
Pros & cons of hard stop-loss systems
- Sunk-cost mechanically eliminated
- Loss size capped (e.g. −20%)
- Capital redeployed
- Tax-loss can be harvested
- Volatile quality stocks (Tesla, NVIDIA) need wide stops
- Market crashes deliver 30% drawdowns even on good stocks
- Mechanical stops can knock you out of compounders
Recommendation: re-entry test as universal tool, hard stops only on speculative names, wide stops (20-30%) on core positions.
Common questions
Is sunk-cost the same as loss aversion?
Related, not identical. Loss aversion is the valuation (pain vs joy). Sunk-cost fallacy is the wrongful inclusion of past costs into decision logic. Loss aversion produces sunk-cost behaviour — sunk-cost is the direct logical consequence.
What about “buy & hold strategy”?
Buy & Hold isn’t sunk-cost-driven if it’s a deliberate, pre-defined strategy. Sunk-cost is reactive holding without clear logic — “because I already lost $5,000”. True buy & hold says: “because I believe in 20 years of compounding”.
How often should I run the re-entry test?
Recommendation: monthly for any position > 5% portfolio share, quarterly for smaller. Spot check before any major market event (earnings, macro events).
What if the loss exceeds my tax-loss bucket?
US: $3,000 ordinary-income offset per year, surplus carries forward. Germany: stock losses only offset stock gains; surplus carries forward indefinitely. Don’t worry — the loss isn’t lost, just deferred.
What if the stock came from a family member?
Inheritance is sunk-cost trap plus endowment effect plus emotional attachment. Mandatory question: would I buy today? If no → sell. Often you can honour the deceased by deploying capital into a position that fits your own strategy.
Loss realisation in a crisis — counterproductive?
Selling during a crash isn’t anti-sunk-cost, it’s panic selling. Important: run the re-entry test — if the thesis is intact, hold or add. If the thesis is genuinely broken by the crisis, then sell. Sunk-cost fallacy doesn’t mean “always sell in a crash”.
Related Hubs: Investor Glossary | Legendary Quotes
Try TradingView Free for 30 Days
Plus get a $15 discount on your first subscription through this link.