Disposition Effect — Why Your Rebalancing Is Systematically Off
You hold a stock that’s up 60%. You sell it. You hold another that’s down 40%. You keep it. Both feel right — and statistically both are exactly the wrong trade. Shefrin and Statman labelled the pattern in 1985 the “disposition effect” and proved: investors realise gains at twice the frequency of losses. It costs 1-2% return per year and torpedoes your rebalancing. Here are the mechanics, the tax logic, and a rule system that flips the bias.
What the disposition effect is
The disposition effect is a direct child of loss aversion. You sell winners because closing a profit position feels good (realised gain). You hold losers because closing a losing position triggers pain (realised loss). It’s asymmetric pain avoidance — not a valuation judgment.
Shefrin/Statman 1985, Odean 1998: investors close gain positions ~50% more frequently than equally aged loss positions. Retail average 1.8 to 2.2x ratio. Pros show the effect attenuated but present (Frazzini 2006).
Why your rebalancing gets destroyed
Classic rebalancing mantra: after large moves, restore allocation to target weights. The disposition effect kills it systematically:
- Winners get trimmed too early: Apple is up 50% in 8 months. Portfolio share went from 5% to 7.5%. You trim back to 5%. But Apple keeps compounding — you capped a 12-month winner too early.
- Losers don’t get topped up: Bayer is down 40%, share dropped from 5% to 3%. Clean rebalance would top up to 5%. But it feels like “throwing money at a loser”, so you stay underweight.
- Conflict with tax logic: Selling winners realises taxes. Holding losers leaves no realised loss to offset. Doubly costly.
Tax-loss harvesting as anti-disposition
Reversing the effect wins both mentally and on tax:
- Realise losers: When the thesis is gone (or at least less certain), realise the loss. In the US, $3,000 net loss can offset ordinary income, surplus carries forward. In Germany the loss enters the stock-loss bucket and offsets stock gains in the same or later years.
- Let winners run: A compounder with intact thesis stays. Trim only if position-size cap is breached (e.g. 10% per single name). Pay tax only when risk-driven.
- Substitute, don’t just hold: Instead of holding Bayer down 40%, sell + buy a sector ETF (Healthcare ETF). Loss realised for tax, market exposure preserved. Mind the wash-sale rule (US 30 days, none in Germany).
- End-Q4 loss harvesting: Standard practice early December: review every loss position for sale rationality. Realising Bayer in early December offsets summer NVIDIA gains.
How much does the disposition effect cost?
| Study | Finding | Return drag |
|---|---|---|
| Odean 1998 (10,000 US accounts) | Sold winners outperform held losers by 4.4% | ~4.4% p.a. opportunity loss |
| Frazzini 2006 (mutual funds) | Even pro managers show the effect | 0.3-1.2% p.a. |
| Grinblatt/Keloharju 2001 (Finland) | Effect independent of education, age, wealth | 1-2% p.a. structural drag |
| BMI investor data 2024 | Gain positions held 3.1× shorter than loss positions | 1-2% p.a. |
Plus the tax optimisation you don’t realise: on a $100k portfolio with $5k loss potential, harvesting can deliver $1,200 in tax savings. Compounded over 10 years that adds up significantly.
Three rule systems against the effect
- Re-entry test per position: Monthly per stock ask: “Would I buy this today?” If yes → hold or add. If no → sell, regardless of gain or loss. This single question neutralises the effect by ~70%.
- Position-size cap with trim trigger: Max 8-10% per single position. If breached, mechanically trim back. Big winners get clipped, but only when systemic risk emerges — not for the dopamine of a realised gain.
- Calendar-triggered loss realisation: First week of December: review all loss positions, sell those without intact thesis. Loss realisation becomes routine, not drama.
Example: disposition vs anti-disposition
Pros & cons of automated anti-disposition
- Loss realisation becomes routine, not drama
- Tax-loss harvesting actually happens
- Winners not reflexively trimmed
- Rebalancing logic stays intact
- Some loss positions are real bargains (Bayer could recover) and selling would be a mistake
- Mechanical loss harvesting can miss recoveries
- Tax-loss bucket has limits — losses above the cap can’t all be absorbed
Solution: not “always realise losses”, but “realise losses when thesis is gone”. The re-entry test is the right filter question.
Common questions
How is the disposition effect different from loss aversion?
Loss aversion is the valuation (pain vs joy). Disposition effect is the behavioural consequence (realisation frequency). Parent → child. Loss aversion produces disposition; disposition is the direct sale-decision impact.
Does a stop-loss help against disposition?
Partially. Pre-trade stop prevents the “hold until recovery” pattern — good. Weakness: tight stops on volatile names knock you out of real compounders. Solution: wide stops (20-30%) on speculative names, re-entry test on core positions.
What about wash-sale rules?
US: 30-day wash-sale rule — sell at a loss, can’t buy substantially identical security 30 days before/after, else loss disallowed. UK: 30-day “bed and breakfasting” rule. Germany: no wash-sale rule. Always check your jurisdiction.
What about ETFs and disposition?
ETFs lack single-stock theses, so the classic effect is dampened. But investors show ETF-allocation disposition: a sector ETF that’s fallen is held longer than one that’s risen. Solution: annual rebalancing to target weights (mechanical, not emotional).
How do I distinguish “intact thesis” from “hope”?
Intact thesis has 3 sentences of written rationale with verifiable numbers (cashflows growing, market share up, margin stable). Hope has one sentence without numbers (“it’ll come back at some point”). Without 3 verifiable sentences, it’s hope — sell.
Does tax-loss harvesting make sense for savings-plan investors?
Yes, but check effort vs return. Below ~$50k portfolio, limited utility. Above $100k, $1-2k tax savings per year is realistic — worth the effort.
Related Hubs: Investor Glossary | Legendary Quotes
Try TradingView Free for 30 Days
Plus get a $15 discount on your first subscription through this link.